Consistent Global States of Distributed Systems:

Fundamental Concepts and Mechanisms

Özalp Babaoğlu Keith Marzullo

Technical Report UBLCS-93-1

January 1993

CM Laboratory for Computer Science

University of Bologna
Piazza di Porta S. Donato, 5
40127 Bologna (Italy)
The University of Bologna Laboratory for Computer Science Research Technical Reports are available
via anonymous FTP from the area ftp.cs.unibo.it:/pub/TR/UBLCS in compressed PostScript
format. Abstracts are available from the same host in the directory /pub/TR/ABSTRACTS in plain
text format. All local authors can be reached via e-mail at the address [email protected].

UBLCS Technical Report Series

92-1 Mapping Parallel Computations onto Distributed Systems in Paralex, by Ö. Babaoğlu, L. Alvisi, A.
Amoroso and R. Davoli, January 1992.
92-2 Parallel Scientific Computing in Distributed Systems: The Paralex Approach, by L. Alvisi, A.
Amoroso, Ö. Babaoğlu, A. Baronio, R. Davoli and L. A. Giachini, February 1992.
92-3 Run-time Support for Dynamic Load Balancing and Debugging in Paralex, by Ö. Babaoğlu, L.
Alvisi, S. Amoroso, R. Davoli, L. A. Giachini, September 1992.
92-4 Paralex: An Environment for Parallel Programming in Distributed Systems, by Ö. Babaoğlu, L.
Alvisi, S. Amoroso, R. Davoli, L. A. Giachini, October 1992.
93-1 Consistent Global States of Distributed Systems: Fundamental Concepts and Mechanism, by Ö.
Babaoğlu and K. Marzullo, January 1993.
93-2 Understanding Non-Blocking Atomic Commitment, by Ö. Babaoğlu and S. Toueg, January 1993.
Consistent Global States of Distributed Systems:
Fundamental Concepts and Mechanisms

Özalp Babaoğlu1 Keith Marzullo 2

Technical Report UBLCS-93-1

January 1993

Abstract
Many important problems in distributed computing admit solutions that contain a phase where
some global property needs to be detected. This subproblem can be seen as an instance of the
Global Predicate Evaluation (GPE) problem where the objective is to establish the truth of a
Boolean expression whose variables may refer to the global system state. Given the uncertainties
in asynchronous distributed systems that arise from communication delays and relative speeds of
computations, the formulation and solution of GPE reveal most of the subtleties in global reasoning
with imperfect information. In this paper, we use GPE as a canonical problem in order to survey
concepts and mechanisms that are useful in understanding global states of distributed computations.
We illustrate the utility of the developed techniques by examining distributed deadlock detection and
distributed debugging as two instances of GPE.

1. Department of Mathematics, University of Bologna, Piazza Porta S. Donato 5, 40127 Bologna, Italy. This
author was supported in part by the Commission of European Communities under ESPRIT Programme Basic
Research Project 6360 (BROADCAST), Hewlett-Packard of Italy and the Italian Ministry of University, Research
and Technology.
2. Department of Computer Science, 4130 Upson Hall, Cornell University, Ithaca, New York 14853 USA. This
author was supported in part by the Defense Advanced Research Projects Agency (DoD) under NASA Ames
grant number NAG 2–593, and by grants from IBM and Siemens. The views, opinions, and findings contained
in this report are those of the authors and should not be construed as an official Department of Defense position,
policy, or decision.

1
 1 Introduction

1 Introduction
A large class of problems in distributed computing can be cast as executing some notification
or reaction when the state of the system satisfies a particular condition. Examples of such
problems include monitoring and debugging, detection of particular states such as deadlock
and termination, and dynamic adaptation of a program’s configuration such as for load
balancing. Thus, the ability to construct a global state and evaluate a predicate over such a
state constitutes the core of solutions to many problems in distributed computing.
The global state of a distributed system is the union of the states of the individual
processes. Given that the processes of a distributed system do not share memory but instead
communicate solely through the exchange of messages, a process that wishes to construct
a global state must infer the remote components of that state through message exchanges.
Thus, a fundamental problem in distributed computing is to ensure that a global state
constructed in this manner is meaningful.
In asynchronous distributed systems, a global state obtained through remote observa-
tions could be obsolete, incomplete, or inconsistent. Informally, a global state is inconsistent
if it could never have been constructed by an idealized observer that is external to the system.
It should be clear that uncertainties in message delays and in relative speeds at which local
computations proceed prevent a process from drawing conclusions about the instantaneous
global state of the system to which it belongs. While simply increasing the frequency of com-
munication may be effective in making local views of a global state more current and more
complete, it is not sufficient for guaranteeing that the global state is consistent. Ensuring the
consistency of a constructed global state requires us to reason about both the order in which
messages are observed by a process as well as the information contained in the messages.
For a large class of problems, consistency turns out to be an appropriate formalization of the
notion that global reasoning with local information is “meaningful”.
Another source of difficulty in distributed systems arises when separate processes
independently construct global states. The variability in message delays could lead to these
separate processes constructing different global states for the same computation. Even
though each such global state may be consistent and the processes may be evaluating the
same predicate, the different processes may execute conflicting reactions. This “relativistic
effect” is inherent to all distributed computations and limits the class of system properties
that can be effectively detected.
In this paper, we formalize and expand the above concepts in the context of an abstract
problem called Global Predicate Evaluation (GPE). The goal of GPE is to determine whether
the global state of the system satisfies some predicate . Global predicates are constructed
so as to encode system properties of interest in terms of state variables. Examples of dis-
tributed system problems where the relevant properties can be encoded as global predicates
include deadlock detection, termination detection, token loss detection, unreachable storage
(garbage) collection, checkpointing and restarting, debugging, and in general, monitoring
and reconfiguration. In this sense, a solution to GPE can be seen as the core of a generic
solution for all these problems; what remains to be done is the formulation of the appropri-
ate predicate and the construction of reactions or notifications to be executed when the
predicate is satisfied.
We begin by defining a formal model for asynchronous distributed systems and
distributed computations. We then examine two different strategies for solving GPE. The
first strategy, introduced in Section 5, and refined in Section 13, is based on a monitor process

UBLCS-93-1 2
 2 Asynchronous Distributed Systems

that actively interrogates the rest of the system in order to construct the global state. In
Section 6 we give a formal definition for consistency of global states. The alternative strategy,
discussed in Section 7, has the monitor passively observe the system in order to construct its
global states. Sections 8 – 13 introduce a series of concepts and mechanisms necessary for
making the two strategies work efficiently. In Section 14 we identify properties that global
predicates must satisfy in order to solve practical problems using GPE. In Section 15 we
address the issue of multiple monitors observing the same computation. We illustrate the
utility of the underlying concepts and mechanisms by applying them to deadlock detection
and to debugging in distributed systems.

2 Asynchronous Distributed Systems

A distributed system is a collection of sequential processes 1 2 . . . and a network capa-
ble of implementing unidirectional communication channels between pairs of processes for
message exchange. Channels are reliable but may deliver messages out of order. We assume
that every process can communicate with every other process, perhaps through intermediary
processes. In other words, the communication network is assumed to be strongly connected
(but not necessarily completely connected).
In defining the properties of a distributed system, we would like to make the weakest
set of assumptions possible. Doing so will enable us to establish upper bounds on the costs
of solving problems in distributed systems. More specifically, if there exists a solution to a
problem in this weakest model with some cost , then there is a solution to the same problem
with a cost no greater than in any distributed system.
The weakest possible model for a distributed system is called an asynchronous sys-
tem and is characterized by the following properties: there exist no bounds on the relative
speeds of processes and there exist no bounds on message delays. Asynchronous systems
rule out the possibility of processes maintaining synchronized local clocks [18,8] or reason-
ing based on global real-time. Communication remains the only possible mechanism for
synchronization in such systems.
In addition to their theoretical interest as noted above, asynchronous distributed
systems may also be realistic models for actual systems. It is often the case that phys-
ical components from which we construct distributed systems are synchronous. In other
words, the relative speeds of processors and message delays over network links making up
a distributed system can be bounded. When, however, layers of software are introduced
to multiplex these physical resources to create abstractions such as processes and (reliable)
communication channels, the resulting system may be better characterized as asynchronous.

3 Distributed Computations
Informally, a distributed computation describes the execution of a distributed program by
a collection of processes. The activity of each sequential process is modeled as executing a
sequence of events. An event may be either internal to a process and cause only a local state
change, or it may involve communication with another process. Without loss of generality,
we assume that communication is accomplished through the events and
that match based on the message identifier . In other words, even if several processes

UBLCS-93-1 3
 3 Distributed Computations

send the same data value to the same process, the messages themselves will be unique.3
Informally, the event enqueues message on an outgoing channel for transmission
to the destination process. The event , on the other hand, corresponds to the
act of dequeuing message from an incoming channel at the destination process. Clearly,
for event to occur at process , message must have arrived at and must
have declared its willingness to receive a message. Otherwise, either the message is delayed
(because the process is not ready) or the process is delayed (because the message has not
arrived).
Note that this “message passing” view of communication at the event level may be
quite different from those of higher system layers. Remote communication at the program-
ming language level may be accomplished through any number of paradigms including
remote procedure calls [4], broadcasts [12], distributed transactions [19], distributed ob-
jects [20] or distributed shared memory [21]. At the level we observe distributed computa-
tions, however, all such high-level communication boil down to generating matching send
and receive events at pairs of processes.
The local history of process during the computation is a (possibly infinite) sequence
of events 1 2 . . .. This labeling of the events of process where 1 is the first event exe-
2
cuted, is the second event executed, etc. is called the canonical enumeration and corresponds
1 2
to the total order imposed by the sequential execution on the local events. Let ...
denote an initial prefix of local history containing the first events. We define to be the 0

empty sequence. The global history of the computation is a set 1 containing

all of its events.4
Note that a global history does not specify any relative timing between events. In an
asynchronous distributed system where no global time frame exists, events of a computation
can be ordered only based on the notion of “cause-and-effect”. In other words, two events
are constrained to occur in a certain order only if the occurrence of the first may affect the
outcome of the second. This in turn implies that information flows from the first event to the
second. In an asynchronous system, information may flow from one event to another either
because the two events are of the same process, and thus may access the same local state, or
because the two events are of different processes and they correspond to the exchange of a
message. We can formalize these ideas by defining a binary relation defined over events
such that [17]:
1. If and , then ,
2. If and , then ,
3. If and , then .
As defined, this relation effectively captures our intuitive notion of “cause-and-effect”
in that if and only if causally precedes .5 Note that only in the case of matching send-
receive events is the cause-and-effect relationship certain. In general, the only conclusion
that can be drawn from is that the mere occurrence of and its outcome may have been
influenced by event .

3. For finite computations, this can be easily accomplished by adding the process index and a sequence number
to the data value to construct the message identifier.
4. Sometimes we are interested in local histories as sets rather than sequences of events. Since all events of a
computation have unique labels in the canonical enumeration, as a set contains exactly the same events as
as a sequence. We use the same symbol to denote both when the appropriate interpretation is clear from context.
5. While “ may causally affect ”, or, “ occurs in the causal context of ” [28] are equivalent interpretations
of this relation, we prefer not to interpret it as “ happens before ” [17] because of the real-time connotation.

UBLCS-93-1 4
 4 Global States, Cuts and Runs

1 2 3 4 5 6
1 1 1 1 1 1
1

2
2
2
1 3
2 2

3
1 2 3 4 5 6
3 3 3 3 3 3

Figure 1. Space-Time Diagram Representation of a Distributed Computation

Certain events of the global history may be causally unrelated. In other words, it is
possible that for some and , neither nor . We call such events concurrent and
write .
Formally, a distributed computation is a partially ordered set (poset) defined by the pair
. Note that all events are labeled with their canonical enumeration, and in the case
of communication events, they also contain the unique message identifier. Thus, the total
ordering of events for each process as well as the send-receive matchings are implicit in .
It is common to depict distributed computations using an equivalent graphical repre-
sentation called a space-time diagram. Figure 1 illustrates such a diagram where the horizontal
lines represent execution of processes, with time progressing from left to right. An arrow
from one process to another represents a message being sent, with the send event at the
base of the arrow and the corresponding receive event at the head of the arrow. Internal
events have no arrows associated with them. Given this graphical representation, it is easy
to verify if two events are causally related: if a path can be traced from one event to the other
proceeding left-to-right along the horizontal lines and in the sense of the arrows, then they
are related; otherwise they are concurrent. For example, in the figure 12 6 but 2 6 .
3 2 3

4 Global States, Cuts and Runs

Let denote the local state of process immediately after having executed event and let
0 be its initial state before any events are executed. In general, the local state of a process

may include information such as the values of local variables and the sequences of messages
sent and received over the various channels incident to the process. The global state of a
distributed computation is an -tuple of local states 1 ... , one for each process.6

6. We can define global states without referring to channel states since they can always be encoded as part of
the process local states. We discuss explicit representation of channel states in Section 13.

UBLCS-93-1 5
 5 Monitoring Distributed Computations

1 2 3 4 5 6
1 1 1 1 1 1
1

2
2
2
1 3
2 2

3
1 2 3 4 5 6
3 3 3 3 3 3

Figure 2. Cuts of a Distributed Computation

A cut of a distributed computation is a subset of its global history and contains an initial
prefix of each of the local histories. We can specify such a cut 1
1
through the
tuple of natural numbers 1 . . . corresponding to the index of the last event included
for each process. The set of last events 11 . . . included in cut 1 . . . is called the
frontier of the cut. Clearly, each cut defined by 1 . . . has a corresponding global state
which is 11 . . . .
As shown in Figure 2, a cut has a natural graphical interpretation as a partitioning
of the space-time diagram along the time axis. The figure illustrates two cuts and
corresponding to the tuples 5 2 4 and 3 2 6 , respectively.
Even though a distributed computation is a partially ordered set of events, in an actual
execution, all events, including those at different processes, occur in some total order.7 To
be able to reason about executions in distributed systems, we introduce the notion of a run.
A run of a distributed computation is total ordering that includes all of the events in the
global history and that is consistent with each local history. In other words, for each process
, the events of appear in in the same order that they appear in . Note that a run need
not correspond to any possible execution and a single distributed computation may have
many runs, each corresponding to a different execution.

5 Monitoring Distributed Computations

Given the above notation and terminology, GPE can be stated as evaluating a predicate
that is a function of the global state of a distributed system. For the time being, we will
assume that a single process called the monitor is responsible for evaluating . Let 0 be

7. If two events actually do occur at the same real-time, we can arbitrarily say that the event of the process with
the smaller index occurs before the event of the larger-index process.

UBLCS-93-1 6
 5 Monitoring Distributed Computations

this process which may be one of 1 . . . or may be external to the computation (but not
the system). In this special case, where there is a single monitor, solving GPE reduces to
0 constructing a global state of the computation (to which is applied). For simplicity
of exposition, we assume that events executed on behalf of monitoring are external to the
underlying computation and do not alter the canonical enumeration of its events.
In the first strategy we pursue for constructing global states, the monitor 0 takes
on an active role and sends each process a “state enquiry” message. Upon the receipt of
such a message, replies with its current local state . When all processes have replied,
0 can construct the global state 1 ... . Note that the positions in the process local
histories that state enquiry messages are received effectively defines a cut. The global state
constructed by 0 is the one corresponding to this cut.
Given that the monitor process is part of the distributed system and is subject to the
same uncertainties as any other process, the simple-minded approach sketched above may
lead to predicate values that are not meaningful. To illustrate the problems that can arise,
consider a distributed system composed of servers providing remote services and clients that
invoke them. In order to satisfy a request, a server may invoke other services (and thus
act as a client). Clients and servers interact through remote procedure calls—after issuing a
request for service, the client remains blocked until it receives the response from the server.
The computation depicted in Figure 1 could correspond to this interaction if we interpret
messages labeled req as requests for service and those labeled resp as responses. Clearly,
such a system can deadlock. Thus, it is important to be able to detect when the state of this
system includes deadlocked processes.
One possibility for detecting deadlocks in the above system is as follows. Server
processes maintain local states containing the names of clients from which they received
requests but to which they have not yet responded. The relevant aspects of the global state
of this system can be summarized through a waits-for graph (WFG ) where the nodes
correspond to processes and the edges model blocking. In this graph, an edge is drawn from
node to node if has received a request from to which it has not yet responded. Note
that WFG can be constructed solely on the basis of local states. It is well known that a cycle
in WFG is a sufficient condition to characterize deadlock in this system [11]. The nodes
of the cycle are exactly those processes involved in the deadlock. Thus, the predicate
“WFG contains a cycle” is one possibility for deadlock detection.8
Let us see what might happen if process 0 monitors the computation of Figure 1 as
outlined above. Suppose that the state enquiry messages of 0 are received by the three
application processes at the points corresponding to cut of Figure 2. In other words,
3 2 6
processes 1 , 2 and 3 report local states 1 , 2 and 3 , respectively. The WFG constructed
by 0 for this global state will have edges 1 3 , 2 1 and 3 2 forming a cycle. Thus, 0
will report a deadlock involving all three processes.
An omniscient external observer of the computation in Figure 1, on the other hand,
would conclude that at no time is the system in a deadlock state. The condition detected
by 0 above is called a ghost deadlock in that it is fictitious. While every cut of a distributed
computation corresponds to a global state, only certain cuts correspond to global states that
could have taken place during a run. Cut of Figure 2 represents such a global state. On

8. Note that defined as a cycle in WFG characterizes a stronger condition than deadlock in the sense that
implies deadlock but not vice versa. If, however, processes can receive and record requests while being blocked,
then a deadlocked system will eventually satisfy .

UBLCS-93-1 7
 6 Consistency

the other hand, cut constructed by 0 corresponds to a global state that could never occur
since process 3 is in a state reflecting the receipt of a request from process 1 that 1 has no
record of having sent. Predicates applied to cuts such as can lead to incorrect conclusions
about the system state.
We return to solving the GPE problem through active monitoring of distributed com-
putations in Section 13 after understanding why the above approach failed.

6 Consistency
Causal precedence happens to be the appropriate formalism for distinguishing the two
classes of cuts exemplified by and . A cut is consistent if for all events and

In other words, a consistent cut is left closed under the causal precedence relation. In its
graphical representation, verifying the consistency of a cut becomes easy: if all arrows that
intersect the cut have their bases to the left and heads to the right of it, then the cut is
consistent; otherwise it is inconsistent. According to this definition, cut of Figure 2 is
consistent while cut is inconsistent. A consistent global state is one corresponding to a
consistent cut. These definitions correspond exactly to the intuition that consistent global
states are those that could occur during a run in the sense that they could be constructed
by an idealized observer external to the system. We can now explain the ghost deadlock
detected by 0 in the previous section as resulting from the evaluation of in an inconsistent
global state.
Consistent cuts (and consistent global states) are fundamental towards understanding
asynchronous distributed computing. Just as a scalar time value denotes a particular instant
during a sequential computation, the frontier of a consistent cut establishes an “instant”
during a distributed computation. Similarly, notions such as “before” and “after” that are
defined with respect to a given time in sequential systems have to be interpreted with respect
to consistent cuts in distributed system: an event is before (after) a cut if is to the left
(right) of the frontier of .
Predicate values are meaningful only when evaluated in consistent global states since
these characterize exactly the states that could have taken place during an execution. A
run is said to be consistent if for all events, implies that appears before in . In
other words, the total order imposed by on the events is an extension of the partial order
defined by causal precedence. It is easy to see that a run 1 2 . . . results in a sequence
0 1 2 0
of global states . . . where denotes the initial global state 10 . . . 0 . If the run is
consistent, then the global states in the sequence will all be consistent as well. We will use
the term “run” to refer to both the sequence of events and the sequence of resulting global
states. Each (consistent) global state of the run is obtained from the previous state 1

by some process executing the single event . For two such (consistent) global states of run
, we say that 1 leads to in . Let denote the transitive closure of the leads-to
relation in a given run . We say that is reachable from in run if and only if .
We drop the run subscript if there exists some run in which is reachable from .
The set of all consistent global states of a computation along with the leads-to relation
defines a lattice. The lattice consists of orthogonal axes, with one axis for each process. Let
1 ... be a shorthand for the global state 1 1 . . . and let 1 be its level. Figure 3

UBLCS-93-1 8
 6 Consistency

00

10 01

11 02

21 12 03

31 22 13 04

41 32 23 14

42 33 24
1 2 3 4 5 6
1 1 1 1 1 1
43 34 1

53 44 35

63 54 45

64 55
2
1 2 3 4 5
65 2 2 2 2 2

Figure 3. A Distributed Computation and the Lattice of its Global States

UBLCS-93-1 9
 7 Observing Distributed Computations

illustrates a distributed computation of two processes and the corresponding global state
lattice. Note that every global state is reachable from the initial global state 00 . A path in
the lattice is a sequence of global states of increasing level (in the figure, downwards) where
the level between any two successive elements differs by one. Each such path corresponds
to a consistent run of the computation. The run is said to “pass through” the global states
included in the path. For the example illustrated in Figure 3, one possible run may pass
through the sequence of global states
00 01 11 21 31 32 42 43 44 54 64 65

Note that one might be tempted to identify the run corresponding to the actual execu-
tion of the computation. As we argued earlier, in an asynchronous distributed system, this
is impossible to achieve from within the system. Only an omniscient external observer will
be able to identify the sequence of global states that the execution passed through.

7 Observing Distributed Computations

Let us consider an alternative strategy for the monitor process 0 in constructing global states
to be used in predicate evaluation based on a reactive architecture [13]. In this approach, 0
will assume a passive role in that it will not send any messages of its own. The application
processes, however, will be modified slightly so that whenever they execute an event, they
notify 0 by sending it a message describing the event.9 As before, we assume that monitoring
does not generate any new events in that the send to 0 for notification coincides with the
event it is notifying. In this manner, the monitor process constructs an observation of the
underlying distributed computation as the sequence of events corresponding to the order in
which the notification messages arrive [14].
We note certain properties of observations as constructed above. First, due to the
variability of the notification message delays, a single run of a distributed computation may
have different observations at different monitors. This is the so-called “relativistic effect”
of distributed computing to which we return in Section 15. Second, an observation can
correspond to a consistent run, an inconsistent run or no run at all since events from the
same process may be observed in an order different from their local history. A consistent
observation is one that corresponds to a consistent run. To illustrate these points, consider
the following (consistent) run of the computation in Figure 1:
1 1 2 1 3 4 2 2 5 3 4 5 6 3 6
3 1 3 2 3 3 2 1 3 1 1 1 3 2 1

All of the following are possible observations of :

1 1 1 2 4 2 2 3 3 4 5
1 2 1 3 3 3 1 2 3 1 1 3 ...
1 1 1 2 2 3 4 3 2 5 6
2 1 3 2 3 1 3 3 1 2 3 3 ...
1 1 1 2 2 3 3 4 4 2 5
3 3 2 1 1 3 3 1 3 1 2 1 ...

9. In general, the application processes need to inform 0 only when they execute an event that is relevant to .
A local event is said to be relevant to predicate if the value of evaluated in a global state . . . . . . could
1
be different from that evaluated in . . . . . . . For example, in the client-server computation of Figure 1, the
only events relevant to deadlock detection are the sending/receiving of request and response messages since
only these can change the state of the WFG .

UBLCS-93-1 10
 7 Observing Distributed Computations

Given our asynchronous distributed system model where communication channels

need not preserve message order, any permutation of run is a possible observation of it.
Not all observations, however, need be meaningful with respect to the run that produced
them. For example, among those indicated above, observation 1 does not even correspond
to a run since events of process 3 do not represent an initial prefix of its local history ( 43
appears before event 33 ). Observation 2 , on the hand, corresponds to an inconsistent run.
In fact, the global state constructed by 0 at the end of observation 2 would be 13 22 36 ,
which is exactly the global state defined by cut of Figure 2 resulting in the detection of a
ghost deadlock. Finally, 3 is a consistent observation and leads to the same global state as
that of cut in Figure 2.
It is the possibility of messages being reordered by channels that leads to undesirable
observations such as 1 . We can restore order to messages between pairs of processes
by defining a delivery rule for deciding when received messages are to be presented to the
application process. We call the primitive invoked by the application deliver to distinguish it
from receive, which remains hidden within the delivery rule and does not appear in the local
history of the process.
Communication from process to is said to satisfy First-In-First-Out (FIFO) delivery
if for all messages and
10
FIFO Delivery:
In other words, FIFO delivery prevents one message overtaking an earlier message
sent by the same process. For each source-destination pair, FIFO delivery can be implemented
over non-FIFO channels simply by having the source process add a sequence number to its
messages and by using a delivery rule at the destination that presents messages in an order
corresponding to the sequence numbers. While FIFO delivery is sufficient to guarantee that
observations correspond to runs, it is not sufficient to guarantee consistent observations.
To pursue this approach for solving the GPE problem where is evaluated in global states
constructed from observations, we need to devise a mechanism that ensures their consistency.
We proceed by devising a simple mechanism and refining it as we relax assumptions.
Initially, assume that all processes have access to a global real-time clock and that all message
delays are bounded by . This is clearly not an asynchronous system but will serve as a
starting point. Let RC denote the value of the global clock when event is executed.
When a process notifies 0 of some local event , it includes RC in the notification message
as a timestamp. The delivery rule employed by 0 is the following:
DR1: At time , deliver all received messages with timestamps up to in increasing
timestamp order.
To see why an observation constructed by 0 using DR1 is guaranteed to be con-
sistent, first note that an event is observed before event if and only if RC RC .11
This is true because messages are delivered in increasing timestamp order and delivering
only messages with timestamps up to time ensures that no future message can arrive
with a timestamp smaller than any of the messages already delivered. Since the observation
coincides with the delivery order, is consistent if and only if
Clock Condition: RC RC .
This condition is certainly satisfied when timestamps are generated using the global
real-time clock. As it turns out, the clock condition can be satisfied without any assumptions—

10. Subscripts identify the process executing the event.

11. Again, we can break ties due to simultaneous events based on process indexes.

UBLCS-93-1 11
 8 Logical Clocks

Figure 4. Logical Clocks

in an asynchronous system.

8 Logical Clocks
In an asynchronous system where no global real-time clock can exist, we can devise a simple
clock mechanism for “timing” such that event orderings based on increasing clock values
are guaranteed to be consistent with causal precedence. In other words, the clock condition
can be satisfied in an asynchronous system. For many applications, including the one above,
any mechanism satisfying the clock condition can be shown to be sufficient for using the
values produced by it as if they were produced by a global real-time clock [27].
The mechanism works as follows. Each process maintains a local variable LC called
its logical clock that maps events to the positive natural numbers [17]. The value of the logical
clock when event is executed by process is denoted LC . We use LC to refer to the
current logical clock value of a process that is implicit from context. Each message that is
sent contains a timestamp TS which is the logical clock value associated with the sending
event. Before any events are executed, all processes initialize their logical clocks to zero. The
following update rules define how the logical clock is modified by with the occurrence of
each new event :

LC 1 if is an internal or send event

LC :
max LC TS 1 if

In other words, when a receive event is executed, the logical clock is updated to
be greater than both the previous local value and the timestamp of the incoming message.
Otherwise (i.e., an internal or send event is executed), the logical clock is simply incremented.
Figure 4 illustrates the logical clock values that result when these rules are applied to the
computation of Figure 1.

UBLCS-93-1 12
 8 Logical Clocks

Note that the above construction produces logical clock values that are increasing
with respect to causal precedence. It is easy to verify that for any two events where ,
the logical clocks associated with them are such that LC LC . Thus, logical clocks
satisfy the clock condition of the previous section.12
Now let us return to the goal at hand, which is constructing consistent observations
in asynchronous systems. In the previous section, we argued that delivery rule DR1 lead
to consistent observations as long as timestamps satisfied the clock condition. We have
just shown that logical clocks indeed satisfy the clock condition and are realizable in asyn-
chronous systems. Thus, we should be able to use logical clocks to construct consistent
observations in asynchronous systems. Uses of logical clocks in many other contexts are
discussed in [29].
Consider a delivery rule where those messages that are delivered, are delivered in
increasing (logical clock) timestamp order, with ties being broken as usual based on process
index. Applying this rule to the example of Figure 4, 0 would construct the observation
1 1 1 2 2 3 3 4 4 2 5 5 3 6 6
1 2 3 1 3 3 1 3 1 2 3 1 2 1 3
which is indeed consistent. Unfortunately, the delivery rule as stated lacks liveness since,
without a bound on message delays (and a real-time clock to measure it), no message will
ever be delivered for fear of receiving a later message with a smaller timestamp. This is
because logical clocks, when used as a timing mechanism, lack what we call the gap-detection
property:
Gap-Detection: Given two events and along with their clock values LC and LC
where LC LC , determine whether some other event exists such that LC
LC LC .
It is this property that is needed to guarantee liveness for the delivery rule and can
be achieved with logical clocks in an asynchronous system only if we exploit information
in addition to the clock values. One possibility is based on using FIFO communication
between all processes and 0 . As usual, all messages (including those sent to 0 ) carry the
logical clock value of the send event as a timestamp. Since each logical clock is monotone
increasing and FIFO delivery preserves order among messages sent by a single process,
when 0 receives a message from process with timestamp TS , it is certain that no
other message can arrive from such that TS TS . A message received by
process is called stable if no future messages with timestamps smaller than TS can be
received by . Given FIFO communication between all processes and 0 , stability of message
at 0 can be guaranteed when 0 has received at least one message from all other processes
with a timestamp greater than TS . This idea leads to the following delivery rule for
constructing consistent observations when logical clocks are used for timestamps:
DR2: Deliver all received messages that are stable at 0 in increasing timestamp order.13
Note that real-time clocks lack the gap-detection property as well. The assumption,
however, that message delays are bounded by was sufficient to devise a simple stability
check in delivery rule DR1: at time , all received messages with timestamps smaller than
are guaranteed to be stable.

12. Note that logical clocks would continue to satisfy the clock condition with any arbitrary positive integer
(rather than one) as the increment value of the update rules.
13. Even this delivery rule may lack liveness if some processes do not communicate with 0 after a certain point.
Liveness can be obtained by the monitor 0 requesting an acknowledgement from all processes to a periodic
empty message [17]. These acknowledgements serve to “flush out” messages that may have been in the channels.

UBLCS-93-1 13
 9 Causal Delivery

Figure 5. Message Delivery that is FIFO but not Causal

9 Causal Delivery
Recall that FIFO delivery guarantees order to be preserved among messages sent by the same
process. A more general abstraction extends this ordering to all messages that are causally
related, even if they are sent by different processes. The resulting property is called causal
delivery and can be stated as:
Causal Delivery (CD):
for all messages , sending processes and destination process . In other words,
in a system respecting causal delivery, a process cannot known about the existence of a
message (through intermediate messages) any earlier than the event corresponding to the
delivery of that message [31]. Note that having FIFO delivery between all pairs of processes
is not sufficient to guarantee causal delivery. Figure 5 illustrates a computation where all
deliveries (trivially) satisfy FIFO but those of 3 violate CD.
The relevance of causal delivery to the construction of consistent observations is
obvious: if 0 uses a delivery rule satisfying CD, then all of its observations will be consistent.
The correctness of this result is an immediate consequence of the definition of CD, which
coincides with that of a consistent observation. In retrospect, the two delivery rules DR1
and DR2 we developed in the previous sections are instances of CD that work under certain
assumptions. What we seek is an implementation for CD that makes no assumptions beyond
those of asynchronous systems.

10 Constructing the Causal Precedence Relation

Note that we have stated the gap-detection property in terms of clock values. For implement-
ing causal delivery efficiently, what is really needed is an effective procedure for deciding
the following: given events , that are causally related and their clock values, does there
exist some other event such that (i.e., falls in the causal “gap” between and
)?

UBLCS-93-1 14
 10 Constructing the Causal Precedence Relation

By delivering event notification messages in strict increasing timestamp order, rules

DR1 and DR2 assume that RC RC (equivalently, LC LC ) implies . This
is a conservative assumption since timestamps generated using real-time or logical clocks
only guarantee the clock condition, which is this implication in the opposite sense. Given
RC RC (or LC LC ), it may be that causally precedes or that they are
concurrent. What is known for certain is that . Having just received the notification
of event , DR1 and DR2 could unnecessarily delay its delivery even if they could predict
the timestamps of all notifications yet to be received. The delay would be unnecessary if
there existed future notifications with smaller timestamps, but they all happened to be for
events concurrent with .
The observations of the preceding two paragraphs suggest a timing mechanism TC
whereby causal precedence relations between events can be deduced from their timestamps.
We strengthen the clock condition by adding an implication in the other sense to obtain:
Strong Clock Condition: TC TC .
While real-time and logical clocks are consistent with causal precedence, timing mech-
anism TC is said to characterize causal precedence since the entire computation can be re-
constructed from a single observation containing TC as timestamps [9,34]. This is essential
not only for efficient implementation of CD, but also for many other applications (e.g., dis-
tributed debugging discussed in Section 14.2) that require the entire global state lattice rather
than a single path through it.

10.1 Causal Histories

A brute-force approach to satisfying the strong clock condition is to devise a timing mecha-
nism that produces the set of all events that causally precede an event as its “clock” value [34].
We define the causal history of event in distributed computation as the set

In other words, the causal history of event is the smallest consistent cut that includes
. The projection of on process is the set . Figure 6 graphically illustrates
4
the causal history of event 1 as the darkened segments of process local histories leading
4 1 2 3 4 1 1 2 3
towards the event. From the figure, it is easy to see that 1 1 1 1 1 2 3 3 3 .
In principle, maintaining causal histories is simple. Each process initializes local
variable to be the empty set. If is the receive of message by process from , then
is constructed as the union of , the causal history of the previous local event of
and the causal history of the corresponding send event at (included in message as its
timestamp). Otherwise ( is an internal or send event), is the union of and the causal
history of the previous local event.
When causal histories are used as clock values, the strong clock condition can be
satisfied if we interpret clock comparison as set inclusion. From the definition of causal
histories, it follows that

In case , the set inclusion above can be replaced by the simple set membership
test . The unfortunate property of causal histories that renders them impractical is
that they grow rapidly.

UBLCS-93-1 15
 10 Constructing the Causal Precedence Relation

1 2 3 4 5 6
1 1 1 1 1 1
1

2
2
2
1 3
2 2

3
1 2 3 4 5 6
3 3 3 3 3 3

4
Figure 6. Causal History of Event 1

10.2 Vector Clocks

The causal history mechanism proposed in the previous section can be made practical by
periodically pruning segments of history that are known to be common to all events [28].
Alternatively, the causal history can be represented as a fixed-dimensional vector rather
than a set. The resulting growth rate will be logarithmic in the number of events rather than
linear. In what follows, we pursue this approach.
First, note that the projection of causal history on process corresponds to an
initial prefix of the local history of . In other words, for some unique and,
by the canonical enumeration of events, for all . Thus, a single natural
number is sufficient to represent the set . Since 1 , the entire
causal history can be represented by an -dimensional vector VC where for all 1 ,
the th component is defined as

VC if and only if

The resulting mechanism is known as vector clocks and has been discovered indepen-
dently by many researchers in many different contexts (see [34] for a survey). In this scheme,
each process maintains a local vector VC of natural numbers where VC denotes the
vector clock value of when it executes event . As with logical clocks, we use VC to refer to
the current vector clock of a process that is implicit from context. Each process initializes
VC to contain all zeros. All messages contain a timestamp TS which is the vector clock
value of ’s send event. The following update rules define how the vector clock is modified
by with the occurrence of each new event :
VC : VC 1 if is an internal or send event

VC : max VC TS if
VC : VC 1

UBLCS-93-1 16
 10 Constructing the Causal Precedence Relation

Figure 7. Vector Clocks

In other words, an internal or send event simply increments the local component of the
vector clock. A receive event, on the other hand, first updates the vector clock to be greater
than (on a component-by-component basis) both the previous value and the timestamp of
the incoming message, and then increments the local component. Figure 7 illustrates the
vector clocks associated with the events of the distributed computation displayed in Figure 1.
Given the above implementation, the th component of the vector clock of process
has the following operational interpretation for all :

VC number of events of that causally precede event of

On the other hand, VC counts the number of events has executed up to and
including . Equivalently, VC is the ordinal position of event in the canonical
enumeration of ’s events.
From the definition of vector clocks, we can easily derive a collection of useful prop-
erties. Given two -dimensional vectors and of natural numbers, we define the “less
than” relation (written as ) between them as follows

:1 :
This allows us to express the strong clock condition in terms of vector clocks as

Property 1 (Strong Clock Condition)

VC VC

Note that for the above test, it is not necessary to know on which processes the two
events were executed. If this information is available, causal precedence between two events
can be verified through a single scalar comparison.

UBLCS-93-1 17
 10 Constructing the Causal Precedence Relation

Property 2 (Simple Strong Clock Condition) Given event of process and event of process ,
where
VC VC

Note that the condition VC VC is possible and represents the situation

where is the latest event of that causally precedes of (thus must be a send event).
Given this version of the strong clock condition, we obtain a simple test for concur-
rency between events that follows directly from its definition

Property 3 (Concurrent) Given event of process and event of process

VC VC VC VC

Consistency of cuts of a distributed computation can also be easily verified in terms

of vector clocks. Events and are said to be pairwise inconsistent if they cannot belong to
the frontier of the same consistent cut. In terms of vector clocks, this can be expressed as

Property 4 (Pairwise Inconsistent) Event of process is pairwise inconsistent with event of

process , where , if and only if

VC VC VC VC

The two disjuncts characterize exactly the two possibilities for the cut to include at
least one receive event without including its corresponding send event (thus making it
inconsistent). While this property might appear to be equivalent to at first sight,
this is not the case; it is obviously possible for two events to be causally related and yet be
pairwise consistent.
We can then characterize a cut as being consistent if its frontier contains no pairwise
inconsistent events. Given the definition of a cut, it suffices to check pairwise inconsistency
only for those events that are in the frontier of the cut. In terms of vector clocks, the property
becomes

Property 5 (Consistent Cut) A cut defined by 1 ... is consistent if and only if

:1 1 : VC VC

Recall that, for all , the vector clock component VC can be interpreted as the
number of events of that causally precede event of . The component corresponding to
the process itself, on the other hand, counts the total number of events executed by up to
and including . Let # 1 VC 1. Thus, # denotes exactly the number
of events that causally precede in the entire computation.

Property 6 (Counting) Given event of process and its vector clock value VC , the number
of events such that (equivalently, VC VC ) is given by # .

UBLCS-93-1 18
 11 Implementing Causal Delivery with Vector Clocks

Finally, vector clocks supply a weak form of the gap-detection property that logical
and real-time clocks do not. The following property follows directly from the vector clock
update rules and the second form of the Strong Clock Condition. It can be used to determine
if the causal “gap” between two events admits a third event.

Property 7 (Weak Gap-Detection) Given event of process and event of process , if

VC VC for some , then there exists an event such that

The property is “weak” in the sense that, for arbitrary processes and , we
cannot conclude if the three events form a causal chain . For the special case ,
however, the property indeed identifies the sufficient condition to make such a conclusion.

11 Implementing Causal Delivery with Vector Clocks

The weak gap-detection property of the previous section can be exploited to efficiently
implement causal delivery using vector clocks. Assume that processes increment the local
component of their vector clocks only for events that are notified to the monitor. 14 As usual,
each message carries a timestamp TS which is the vector clock value of the event being
notified by . All messages that have been received but not yet delivered by the monitor
process 0 are maintained in a set , initially empty.
A message from process is deliverable as soon as 0 can verify that there are
no other messages (neither in nor in the network) whose sending causally precede that
of . Let be the last message delivered from process , where . Before message
of process can be delivered, 0 must verify two conditions:
1. there is no earlier message from that is undelivered, and
2. there is no undelivered message from such that

The first condition holds if exactly TS 1 messages have already been delivered
from . To verify the second condition, we can use the special case of weak gap-detection
where and , and . Since the two events
and both occur at process , Property 7 can be written as
(Weak Gap-Detection) If TS TS for some , then there exists event
such that

Thus, no undelivered message exists if TS TS , for all . These tests

can be efficiently implemented if 0 maintains an array 1 . . . of counters, initially all
zeros, such that counter contains TS where is the last message that has been
delivered from process . The delivery rule then becomes:

14. Equivalently, processes send a notification message to the monitor for all of their events.

UBLCS-93-1 19
 11 Implementing Causal Delivery with Vector Clocks

Figure 8. Causal Delivery Using Vector Clocks

DR3: (Causal Delivery) Deliver message from process as soon as both of the following
conditions are satisfied

TS 1
TS

When 0 delivers , array is updated by setting to TS .

Figure 8 illustrates the application of this delivery rule by 0 in a sample computation.
The events of 1 and 2 are annotated with the vector clock values while those of 0 indicate
the values of array . Note that the delivery of message is delayed until message has
been received and delivered. Message , on the other hand, can be delivered as soon as it
is received since 0 can verify that all causally preceding messages have been delivered.
At this point, we have a complete reactive-architecture solution to the GPE problem in
asynchronous distributed systems based on passive observations. The steps are as follows.
Processes notify the monitor 0 of relevant events by sending it messages. The monitor
uses a causal delivery rule for the notification messages to construct an observation that
corresponds to a consistent run. The global predicate can be applied to any one of the
global states in the run since each is guaranteed to be consistent. An application of this
solution to deadlock detection is given in Section 14.1.
Causal delivery can be implemented at any process rather than just at the monitor. If
processes communicate exclusively through broadcasts (rather than point-to-point sends),
then delivery rule DR3 remains the appropriate mechanism for achieving causal delivery at
all destinations [3]. The resulting primitive, known as causal broadcast (c.f. Section 15), has
been recognized as an important abstraction for building distributed applications [3,15,28].
If, on the other hand, communication can take place through point-to-point sends, a delivery
rule can be derived based on an extension of vector clocks where each message carries a
timestamp composed of vector clocks (i.e., an matrix) [32,30].

UBLCS-93-1 20
 12 Causal Delivery and Hidden Channels

Figure 9. External Environment as a Hidden Channel

12 Causal Delivery and Hidden Channels

In general, causal delivery allows processes to reason globally about the system using only
local information. For such conclusions to be meaningful, however, we need to restrict our
attention to closed systems—those that constrain all communication to take place within the
boundaries of the computing system. If global reasoning based on causal analysis is applied
to systems that contain so-called hidden channels, incorrect conclusions may be drawn [17].
To illustrate the problem, consider the example taken from [16] and shown in Figure 9.
A physical process is being monitored and controlled by a distributed system consisting of
1 2 and 3 . Process 1 is monitoring the state of a steam pipe and detects its rupture.
The event is notified to the controller process 3 in message . Process 2 is monitoring
the pressure of the same pipe, several meters downstream from 1 . A few seconds after
the rupture of the pipe, 2 detects a drop in the pressure and notifies 3 of the event in
message . Note that from the point of view of explicit communication, messages and
are concurrent. Message arrives at 3 and is delivered without delay since there are
no undelivered messages that causally precede it. As part of its control action, 3 reacts
to the pressure drop by applying more heat to increase the temperature. Some time later,
message arrives reporting the rupture of the pipe. The causal sequence observed by 3 is
pressure drop, apply heat, pipe rupture leading it to conclude that the pipe ruptured due to
the increased temperature. In fact, the opposite is true.
The apparent anomaly is due to the steam pipe which acts as a communication channel
external to the system. The rupture and pressure drop events are indeed causally related even
though it is not captured by the relation. When the pipe is included as a communication
channel, the order in which messages are seen by 3 violates causal delivery. In systems
that are not closed, global reasoning has to be based on totally-ordered observations derived
from global real-time. Since this order is consistent with causal precedence, anomalous
conclusions such as the one above will be avoided.

UBLCS-93-1 21
 13 Distributed Snapshots

13 Distributed Snapshots
In Section 5 we presented a strategy for solving the GPE problem through active monitoring.
In this strategy, 0 requested the states of the other processes and then combined them into
a global state. Such a strategy is often called a “snapshot” protocol, since 0 “takes pictures”
of the individual process states. As we noted, this global state may not be consistent, and
so the monitor may make an incorrect deduction about the system property encoded in the
global predicate.
We will now develop a snapshot protocol that constructs only consistent global states.
The protocol is due to Chandy and Lamport [5], and the development described here is due
to Morgan [26]. For simplicity, we will assume that the channels implement FIFO delivery,
and we omit details of how individual processes return their local states to 0 .
For this protocol, we will introduce the notion of a channel state. For each channel
from to , its state are those messages that has sent to but has not yet received.
Channel states are only a convenience in that each can be inferred from just the local
states and as the set difference between messages sent by to (encoded in )
and messages received by from (encoded in ). In many cases, however, explicit
maintenance of channel state information can lead to more compact process local states and
simpler encoding for the global predicate of interest. For example, when constructing the
waits-for graph in deadlock detection, an edge is drawn from to if is blocked due
to . This relation can be easily obtained from the local process states and channel states:
process is blocked on if records the fact that there is an outstanding request to , and
contains no response messages.
Let IN be the set of processes that have channels connecting them directly to and
OUT be the set of processes to which has a channel. Channels from IN to are
called incoming while channels from to OUT are called outgoing with respect to .
For each execution of the snapshot protocol, a process will record its local state and the
states of its incoming channels ( , for all IN ).

13.1 Snapshot Protocols

We proceed as before by devising a simple protocol based on a strong set of assumptions
and refining the protocol as we relax them. Initially, assume that all processes have access
to a global real-time clock RC, that all message delays are bound by some known value, and
that relative process speeds are bounded.
The first snapshot protocol is based on all processes recording their states at the same
real-time. Process 0 chooses a time far enough in the future in order to guarantee that
a message sent now will be received by all other processes before .15 To facilitate the
recording of channel states, processes include a timestamp in each message indicating when
the message’s send event was executed.

Snapshot Protocol 1
1. Process 0 sends the message “take snapshot at ” to all processes.16

15. Recall that there need not be a channel between all pairs of processes, and so must account for the
possibility of messages being forwarded.
16. For simplicity, we describe the protocols for a single initiation by process 0 . In fact, they can be initiated by
any process, and as long as concurrent initiations can be distinguished, multiple initiations are possible.

UBLCS-93-1 22
 13 Distributed Snapshots

2. When clock RC reads , each process records its local state , sends an empty
message over all of its outgoing channels, and starts recording messages received over
each of its incoming channels. Recording the local state and sending empty messages
are performed before any intervening events are executed on behalf of the underlying
computation.
3. First time receives a message from with timestamp greater than or equal to ,
stops recording messages for that channel and declares as those messages that
have been recorded.
For each IN , the channel state constructed by process contains the set of
messages sent by before and received by after . The empty messages in Step 2
are sent in order to guarantee liveness: 17 process is guaranteed to eventually receive a
message from every incoming channel such that TS .
Being based on real-time, it is easy to see that this protocol constructs a consistent
global state—it constructs a global state that did in fact occur and thus could have been
observed by our idealized external observer. However, it is worth arguing this point a
little more formally. Note that an event is in the consistent cut associated with the
constructed global state if and only if RC . Hence,
RC RC
Since real-time clock RC satisfies the clock condition, the above equation implies that is
a consistent cut. In fact, the clock condition is the only property of RC that is necessary for
to be a consistent cut. Since logical clocks also satisfy the clock condition, we should be
able to substitute logical clocks for real-time clocks in the above protocol.
There are, however, two other properties of synchronous systems used by Snapshot
Protocol 1 that need to be supplied:
The programming construct “when LC do ” doesn’t make sense in the context
of logical clocks since the given value need not be attained by a logical clock.18 For
example, in Figure 4 the logical clock of 3 never attains the value 6, because the
receipt of the message from 1 forces it to jump from 5 to 7. Even if LC does attain
a value of , the programming construct is still problematic. Our rules for updating
logical clocks are based on the occurrence of new events. Thus, at the point LC ,
the event that caused the clock update has been executed rather than the first event of
.
We overcome this problem with the following rules. Suppose contains the statement
“when LC do ”, where generates only internal events or send events. Before
executing an event , process makes the following test:
– If is an internal or send event and LC 2, then executes and then starts
executing .
– If where TS and LC 1, then puts the message back
onto the channel, re-enables for execution, sets LC to 1 and starts executing
.
In Protocol 1, the monitor 0 chooses such that the message “take snapshot at ” is
received by all other processes before time . In an asynchronous system, 0 cannot

17. Our use of empty messages here is not unlike their use in distributed simulation for the purposes of advancing
global virtual time [25].
18. Note that this problem happens to be one aspect of the more general problem of simulating a synchronous
system in an asynchronous one [1].

UBLCS-93-1 23
 13 Distributed Snapshots

compute such a logical clock value. Instead, we assume that there is an integer value
large enough that no logical clock can reach by using the update rules in Section 8.
Assuming the existence of such an requires us to bound both relative process speeds
and message delays, and so we will have to relax it as well. Given the above considerations,
we obtain Snapshot Protocol 2, which differs from Protocol 1 only in its use of logical clocks
in place of the real-time clock.

Snapshot Protocol 2
1. Process 0 sends “take snapshot at ” to all processes and then sets its logical clock to
.
2. When its logical clock reads , process records its local state , sends an empty
message along each outgoing channel, and starts recording messages received over
each of its incoming channels. Recording the local state and sending empty messages
are performed before any intervening events are executed on behalf of the underlying
computation.
3. First time receives a message from with timestamp greater than or equal to ,
stops recording messages for that channel and declares as those messages that
have been recorded.
Channel states are constructed just as in Protocol 1 with playing the role of .
As soon as 0 sets its logical clock to , it will immediately execute Step 2, and the empty
messages sent by it will force the clocks of processes in OUT0 to attain . Since the network
is strongly connected, all of the clocks will eventually attain , and so the protocol is live.
We now remove the need for . Note that, with respect to the above protocol, a
process does nothing between receiving the “take snapshot at ” message and receiving the
first empty message that causes its clock to pass through . Thus, we can eliminate the
message “take snapshot at ” and instead have a process record its state when it receives the
first empty message. Since processes may send empty messages for other purposes, we will
change the message from being empty to one containing a unique value, for example, “take
snapshot”. Furthermore, by making this message contain a unique value, we no longer need
to include timestamps in messages—the message “take snapshot” is the first message that
any process sends after the snapshot time. Doing so removes the last reference to logical
clocks, and so we can eliminate them from our protocol completely.

Snapshot Protocol 3 (Chandy-Lamport [5])

1. Process 0 starts the protocol by sending itself a “take snapshot” message.
2. Let be the process from which receives the “take snapshot” message for the first
time. Upon receiving this message, records its local state and relays the “take
snapshot” message along all of its outgoing channels. No intervening events on behalf
of the underlying computation are executed between these steps. Channel state is
set to empty and starts recording messages received over each of its other incoming
channels.
3. Let be the process from which receives the “take snapshot” message beyond the
first time. Process stops recording messages along the channel from and declares
channel state as those messages that have been recorded.
Since a “take snapshot” message is relayed only upon the first receipt and since the
network is strongly connected, a “take snapshot” message traverses each channel exactly

UBLCS-93-1 24
 13 Distributed Snapshots

1 2 3 4 5 6
1 1 1 1 1 1
1
1

2
1 2 3 4 5
2 2 2 2 2 2

Figure 10. Application of the Chandy-Lamport Snapshot Protocol

once. When process has received a “take snapshot” message from all of its incoming
channels, its contribution to the global state is complete and its participation in the snapshot
protocol ends.
Note that the above protocols can be extended and improved in many ways including
relaxation of the FIFO assumption [23,35] and reduction of the message complexity [24,31].

13.2 Properties of Snapshots

Let be a global state constructed by the Chandy-Lamport distributed snapshot protocol.
In the previous section, we argued that is guaranteed to be consistent. Beyond that,
however, the actual run that the system followed while executing the protocol may not even
pass through . In this section, we show that is not an arbitrary consistent global state,
but one that has useful properties with respect to the run that generated it.
Consider the application of Chandy-Lamport snapshot protocol to the distributed
computation of Figure 3. The composite computation is shown in Figure 10 where solid
arrows indicate messages sent by the underlying computation while dashed arrows indi-
cate “take snapshot” messages sent by the protocol. From the protocol description, the
constructed global state is 23 with 1 2 empty and 2 1 containing message . Let the run
followed by processes 1 and 2 while executing the protocol be
1 1 2 3 2 4 3 4 5 5 6
2 1 1 1 2 1 2 2 1 2 1

or in terms of global states,

00 01 11 21 31 32 42 43 44 54 55 65

Let the global state of this run in which the protocol is initiated be 21 and the global
state in which it terminates be 55 . Note that run does not pass through the constructed
global state 23 . As can be verified by the lattice of Figure 3, however, 21 23 55 in this

example. We now show that this relationship holds in general.

UBLCS-93-1 25
 14 Properties of Global Predicates

Let be the global state in which the snapshot protocol is initiated, be the global
state in which the protocol terminates and be the global state constructed. We will show
that there exists a run such that . Let be the actual run the system
followed while executing the snapshot protocol, and let denote the event when receives
“take snapshot” for the first time, causing to record its state. An event of is a
prerecording event if ; otherwise, it is a postrecording event.
Consider any two adjacent events of such that is a postrecording event and
is a prerecording event.19 We will show that , and so the order of these two events
can be swapped, thereby resulting in another consistent run. If we continue this process
of swapping postrecording, prerecording event pairs, then we will eventually construct a
consistent run in which no prerecording event follows a postrecording event. The global
state associated with the last prerecording event is therefore reachable from and the state
is reachable from it. Finally, we will show that this state is , the state that the snapshot
protocol constructs.
Consider the subsequence of run where is a postrecording event and a
prerecording event. If then the two events cannot be swapped without resulting in an
inconsistent run. For contradiction, assume that . There are two cases to consider:
1. Both events and are from the same process. If this were the case, however, then
by definition would be a postrecording event.
2. Event is a send event of and is the corresponding receive event of . If this
were the case, however, then from the protocol will have sent a “take snapshot”
message to by the time is executed, and since the channel is FIFO, will also be
a postrecording event.
Hence, a postrecording event cannot causally precede a prerecording event and thus
any postrecording, prerecording event pair can be swapped. Let be the run derived
from by swapping such pairs until all postrecording events follow the prerecording events.
We now argue that the global state after the execution of the last prerecording event
in is . By the protocol description and the definition of prerecording, postrecording
events that record local states will record them at point . Furthermore, by the protocol, the
channel states that are recorded are those messages that were sent by prerecording events
and received by postrecording events. By construction, these are exactly those messages in
the channels after the execution of event , and so is the state recorded by the snapshot
protocol.

14 Properties of Global Predicates

We have derived two methods for global predicate evaluation: one based on a monitor
actively constructing snapshots and another one based on a monitor passively observing
runs. The utility of either approach for solving practical distributed systems problems,
however, depends in part on the properties of the predicate that is being evaluated. In this
section, some of these properties are examined.

14.1 Stable Predicates

Let be a consistent global state of a computation constructed through any feasible mech-
anism. Given that communication in a distributed system incurs delays, can only reflect

3 2 4 3
19. Adjacent event pairs 1 2 and 1 2 of run are two such examples.

UBLCS-93-1 26
 14 Properties of Global Predicates

process p(i): 1 i n
var pending: queue of [message, integer] init empty; % pending requests to p(i)
working: boolean init false; % processing a request
m: message; j: integer;
while true do
while working or (size(pending) = 0) do
receive m from p(j); % m set to message, j to its source
case m.type of
request:
pending := pending + [m, j];
response:
[m, j] := NextState(m, j);
working := (m.type = request);
send m to p(j);
esac
od ;
while not working and (size(pending) 0) do
[m, j] := first(pending);
pending := tail(pending);
[m, j] := NextState(m, j);
working := (m.type = request);
send m to p(j)
od
od
end p(i);
Figure 11. Server Process

UBLCS-93-1 27
 14 Properties of Global Predicates

some past state of the system— by the time they are obtained, conclusions drawn about the
system by evaluating predicate in may have no bearing to the present.
Many system properties one wishes to detect, however, have the characteristic that
once they become true, they remain true. Such properties (and their predicates) are called
stable. Examples of stable properties include deadlock, termination, the loss of all tokens,
and unreachability of storage (garbage collection). If is stable, then the monitor process
can strengthen its knowledge about when is satisfied.
As before, let be the global state in which the global state construction protocol is
initiated, be the global state in which the protocol terminates and be the global state
it constructs. Since , if is stable, then the following conclusions are possible
is true in is true in
and
is false in is false in
As an example of detecting a stable property, we return to deadlock in the client-
server system described in Section 5. We assume that there is a bidirectional communication
channel between each pair of processes and each process when acting as a server runs
the same program, shown in Figure 11. The behavior of a process as a client is much
simpler: after sending a request to a server it blocks until the response is received. The
server is modeled as a state machine [33]: it repeatedly receives a message, changes its state,
and optionally sends one or more messages. The function NextState() computes the action a
server next takes: given a message from process , the invocation NextState(m, j) changes
the state of the server and returns the next message to send along with its destination. The
resulting message may be a response to the client’s request or it may be a further request
whose response is needed to service the client’s request. All requests received by a server,
including those received while it is servicing an earlier request, are queued on the FIFO
queue pending, and the server removes and begins servicing the first entry from pending
after it finishes an earlier request by sending the response.
Figures 12 shows the server of Figure 11 with a snapshot protocol embedded in it.
Each server maintains a boolean array blocking that indicates which processes have sent it
requests to which it has not yet responded (this information is also stored in pending, but
we duplicate it in blocking for clarity). When a server first receives a snapshot message, it
sends the contents of blocking to 0 and relays the snapshot message to all other processes.
Subsequent snapshot messages are ignored until has received such messages, one from
each other process.20
The conventional definition of “ waits-for ” is that has sent a request to and
has not yet responded. As in Section 5, we will instead use the weaker definition waits-for
which holds when has received a request from to which it has not yet responded [11].
By structuring the server as a state machine, even requests sent to a deadlocked server will
eventually be received and denoted in blocking. Hence, a system that contains a cycle in the
conventional WFG will eventually contain a cycle in the WFG , and so a deadlock will be
detected eventually. Furthermore, using the WFG instead of the WFG has the advantage
of referencing only the local process states, and so the embedded snapshot protocol need
not record channel states.

20. By the assumption that the network is completely connected, each invocation of the snapshot protocol will
result in exactly snapshot messages to be received by each process.

UBLCS-93-1 28
 14 Properties of Global Predicates

process p(i): 1 i n
var pending: queue of [message, integer] init empty; % pending requests to p(i)
working: boolean init false; % processing a request
blocking: array [1..n] of boolean init false; % blocking[j] = “p(j) is blocked on p(i)”
m: message; j: integer; s: integer init 0;
while true do
while working or (size(pending) = 0) do
receive m from p(j); % m set to message, j to its source
case m.type of
request:
blocking[j] := true;
pending := pending + [m, j];
response:
[m, j] := NextState(m, j);
working := (m.type = request);
send m to p(j);
if (m.type = response) then blocking[j] := false;
snapshot:
if s = 0 then
% this is the first snapshot message
send [type: snapshot, data: blocking] to p(0);
send [type: snapshot] to p(1),...,p(i 1),p(i 1),...,p(n)
s := (s + 1) mod n;
esac
od ;
while not working and (size(pending) 0) do
[m, j] := head(pending);
pending := tail(pending);
[m, j] := NextState(m, j);
working := (m.type = request);
send m to p(j);
if (m.type = response) then blocking[j] := false;
od
od
end p(i);

Figure 12. Deadlock Detection through Snapshots: Server Side

UBLCS-93-1 29
 14 Properties of Global Predicates

process p(0):
var wfg: array [1..n] of array [1..n] of boolean; % wfg[i, j] = p(j) waits-for p(i)
j, k: integer; m: message;
while true do
wait until deadlock is suspected;
send [type: snapshot] to p(1), ..., p(n);
for k := 1 to n do
receive m from p(j);
wfg[j] := m.data;
if (cycle in wfg) then system is deadlocked
od
end p(0);
Figure 13. Deadlock Detection through Snapshots: Monitor Side

Figure 13 shows the code run by the monitor 0 acting as the deadlock detector. This
process periodically starts a snapshot by sending a snapshot message to all other processes.
Then, 0 receives the arrays blocking from each of the processes and uses this data to test
for a cycle in the WFG . This approach has the advantage of generating an additional
message load only when deadlock is suspected. However, the approach also introduces a
latency between the occurrence of the deadlock and detection that depends on how often
the monitor starts a snapshot.
Figures 14 and 15 show the server and monitor code, respectively, of a reactive-
architecture deadlock detector. This solution is much simpler than the snapshot-based
version. In this case, sends a message to 0 whenever receives a request or sends a
response to a client. Monitor 0 uses these notifications to update a WFG in which it tests
for a cycle. The simplicity of this solution is somewhat superficial, however, because the
protocol requires all messages to 0 to be sent using causal delivery order instead of FIFO
order. The only latency between a deadlock’s occurrence and its detection is due to the
delay associated with a notification message, and thus, is typically shorter than that of the
snapshot-based detector.

14.2 Nonstable Predicates

Unfortunately, not all predicates one wishes to detect are stable. For example, when de-
bugging a system one may wish to monitor the lengths of two queues, and notify the user
if the sum of the lengths is larger than some threshold. If both queues are dynamically
changing, then the predicate corresponding to the desired condition is not stable. Detecting
such predicates poses two serious problems.
The first problem is that the condition encoded by the predicate may not persist
long enough for it to be true when the predicate is evaluated. For example, consider the
computation of Figure 16 in which variables and are being maintained by two processes
1 and 2 , respectively. Suppose we are interested in monitoring the condition . There
are seven consistent global states in which the condition holds, yet if the monitor
evaluates after state 54 then the condition will be found not to hold.
The second problem is more disturbing: if a predicate is found to be true by the
monitor, we do not know whether ever held during the actual run. For example, suppose

UBLCS-93-1 30
 14 Properties of Global Predicates

process p(i): 1 i n
var pending: queue of [message, integer] init empty; % pending requests to p(i)
working: boolean init false; % processing a request
m: message; j: integer;
while true do
while working or (size(pending) = 0) do
receive m from p(j); % m set to message, j to its source
case m.type of
request:
send [type: requested, of: i, by: j] to p(0);
pending := pending + [m, j];
response:
[m, j] := NextState(m, j);
working := (m.type = request);
send m to p(j);
if (m.type = response) then
send [type: responded, to: j, by: i] to p(0);
esac
od ;
while not working and (size(pending) 0) do
[m, j] := first(pending);
pending := tail(pending);
[m, j] := NextState(m, j);
working := (m.type = request);
send m to p(j);
if (m.type = response) then
send [type: responded, to: j, by: i] to p(0)
od
od
end p(i);

Figure 14. Deadlock Detection using Reactive Protocol: Server Side

UBLCS-93-1 31
 14 Properties of Global Predicates

process p(0):
var wfg: array [1..n, 1..n] of boolean init false; % wfg[i, j] = “p(j) waits-for p(i)”
m: message; j: integer;
while true do
receive m from p(j); % m set set to message, j to its source
if (m.type = responded) then
wfg[m.by, m.to] := false
else
wfg[m.of, m.by] := true;
if (cycle in wfg) then system is deadlocked
od
end p(0);
Figure 15. Deadlock Detection using Reactive Protocol: Monitor Side

in the same computation the condition being monitored is 2. The only two global
states of Figure 16 satisfying this condition are 31 and 41
. Let a snapshot protocol be
initiated in state 11 of the run
00 01 11 12 22 32 42 43 44 45 55 65

From the result of Section 13.2, we know that the snapshot protocol could construct
either global state 31 or 41 since both are reachable from 11 . Thus, the monitor could
“detect” 2 even though the actual run never goes through a state satisfying the
condition.
It appears that there is very little value in using a snapshot protocol to detect a
nonstable predicate—the predicate may have held even if it is not detected, and even if it is
detected it may have never held. The same problems exist when nonstable predicates are
evaluated over global states constructed from observations of runs: if a nonstable predicate
holds at some state during a consistent observation, then the condition may or may not have
held during the actual run that produced it.
With observations, however, we can extend nonstable global predicates such that they
are meaningful in the face of uncertainty of the run actually followed by the computation.
To do this, the extended predicates must apply to the entire distributed computation rather
than to individual runs or global states of it. There are two choices for defining predicates
over computations [6,22]:
1. Possibly( ): There exists a consistent observation of the computation such that
holds in a global state of .
2. Definitely( ): For every consistent observations of the computation, there exists a
global state of in which holds.
The distributed computation of Figure 16 satisfies both predicates Possibly( 2)
and Definitely( ). As with stable predicates, by the time they are detected, both of these
predicates refer to some past state or states of a run. The predicate “ currently holds” can
also be detected, but to do so will introduce blocking of the underlying computation.
An application of these extended predicates is when a run is observed for purposes
of debugging [6]. For instance, if identifies some erroneous state of the computation, then
Possibly( ) holding indicates a bug, even if it is not observed during an actual run. For

UBLCS-93-1 32
 14 Properties of Global Predicates

00

10 01

11 02
2
21 12 03

31 22 13
: 3 : 4 : 5
1 2 3 4 5 6
41 32 23 1 1 1 1 1 1
1
42 33

43

53 44

2
63 54 45
1 2 3 4 5
2 2 2 2 2
64 55 : 6 : 4 : 2

65 0 10

Figure 16. Global States Satisfying Predicates and 2

UBLCS-93-1 33
 14 Properties of Global Predicates

example, if 2 denotes an erroneous state, then the computation of Figure 16 is

incorrect, since there is no guarantee that the erroneous state will not occur in some run.
The intuitive meanings of Possibly and Definitely could lead one to believe that they
are duals of each other: Definitely being equivalent to Possibly( ) and Possibly
being equivalent to Definitely( ). This is not the case. For example, while it is true
that Definitely holding through a computation does imply Possibly( ) (there must
be an observation in which holds in all of its states), it is possible to have both
Possibly( ) and Definitely( ) hold. Figure 16 illustrates the latter: the computation
satisfies both Possibly( ) and Definitely( ). Furthermore, if predicate is stable,
then Possibly Definitely . The inverse, however, is not true in general.
The choice between detecting whether currently holds versus whether possibly
or definitely held in the past depends on which restrictions confound the debugging pro-
cess more. Detecting a condition that occurred in the past may require program replay or
reverse execution in order to recover the state of interest, which can be very expensive to
provide. Hence, detection in the past is better suited to a post-mortem analysis of a compu-
tation. Detecting the fact that currently holds, on the other hand, requires delaying the
execution of processes, which can be a serious impediment to debugging. By blocking some
processes when the predicate becomes potentially true, we may make the predicate either
more or less likely to occur. For example, a predicate may be less likely to occur if processes
“communicate” using timeouts or some other uncontrolled form of communication. The
latter is a particular problem when processes are multithreaded; that is, consisting of multi-
ple, independently schedulable threads of control which may communicate through shared
memory. In fact, it is rarely practical to monitor such communication when debugging
without hardware or language support.

14.3 Detecting Possibly and Definitely

The algorithms for detecting Possibly( ) and Definitely( ) are based on the lattice of con-
sistent global states associated with the distributed computation. For every global state
in the lattice there exists at least one run that passes through . Hence, if any global state
in the lattice satisfies , then Possibly( ) holds. For example, in the global state lattice of
Figure 16 both 31 and 41 satisfy 2 meaning that Possibly( 2) holds for the
computation. The property Definitely( ) requires all possible runs to pass through a global
state that satisfies . In Figure 16 the global state 43 satisfies . Since this is the only
state with level 7, and all runs must contain a global state of each level, Definitely( )
also holds for the computation.
Figure 17 is a high-level procedure for detecting Possibly( ). The procedure con-
structs the set of global states current with progressively increasing levels (denoted by
). When a member of current satisfies , then the procedure terminates indicating that
Possibly( ) holds. If, however, the procedure constructs the final global state (the global
state in which the computation terminates) and finds that this global state does not satisfy
, then the procedure returns Possibly( ).
In order to implement this procedure, the monitored processes send the portion of
their local states that is referenced in to the monitor 0 . Process 0 maintains sequences
of these local states, one sequence per process, and uses them to construct the global states
of a given level. The basic operation used by the procedure is “current := states of level ”,
and so we must be able to determine when all of the global states of a given level can be
assembled and must be able to effectively assemble them.

UBLCS-93-1 34
 14 Properties of Global Predicates

procedure Possibly( );
var current: set of global states;
: integer;
begin
% Synchronize processes and distribute
send to all processes;
current := global state 0...0 ;
release processes;
:= 0;
0...0
% Invariant: current contains all states of level that are reachable from
while (no state in current satisfies ) do
if current = final global state then return false
:= + 1;
current := states of level
od
return true
end
Figure 17. Algorithm for Detecting Possibly( ).

Let be the sequence of local states, stored in FIFO order, that 0 maintains for
process , where each state in is labeled with the vector timestamp VC . Define
to be the global state with the smallest level containing and to be the
3 31
global state with the largest level containing . For example, in Figure 16, 1
and 3 33 . These states can be computed using Property 5 of vector clocks as
1
follows:
1
1
2 ...
2
: : VC VC
and

1
1
2
2
... :
1
: VC VC VC VC

where is the state in which process terminates.

Global states and bound the levels of the lattice in which
occurs. The minimum level containing is particularly easy to compute: it is the sum of
components of the vector timestamp VC . Thus, 0 can construct the set of states with
level when, for each sequence , the sum of the components of the vector timestamp of
the last element of is at least . For example, if 0 monitors the computation shown in
Figure 16, then 0 can start enumerating level 6 when it has received states 15 and 24 because
any global state containing 15 must have a level of at least 8 (5 3) and any global state
containing 24 must also have a level of at least 8 (4 4). Similarly, process 0 can remove state
from when is greater than the level of the global state . For the example in
2 23 2
Figure 16, 1 , and so 0 can remove 1 from 1 once it has set to 6.
Given the set of states of some level , it is also straightforward (if somewhat costly)
to construct the set of states of level 1: for each state 1 2 of level , one constructs
the global states 1 1 2 , ..., 1 2 1 . Then, Property 5 of vector clocks can be
used to determine which of these global states are consistent. One can be careful and avoid

UBLCS-93-1 35
 14 Properties of Global Predicates

procedure Definitely( );
var current, last: set of global states;
: integer;
begin
% Synchronize processes and distribute
send to all processes;
last := global state 0...0;
release processes;
remove all states in last that satisfy ;
:= 1;
% Invariant: last contains all states of level 1 that are reachable
% from 0...0 without passing through a state satisfying
while (last ) do
current := states of level reachable from a state in last;
remove all states in current that satisfy ;
if current = final global state then return false
:= + 1;
last := current
od
return true
end ;
Figure 18. Algorithm for Detecting Definitely( ).

redundantly constructing the same global state of level 1 from different global states of
level , but the computation can still be costly because the number of states of level grows
1
as (for constant).
Figure 18 gives the high-level algorithm used by the monitoring process 0 to detect
Definitely( ). This algorithm iteratively constructs the set of global states that have a level
and are reachable from the initial global state without passing through a global state that
satisfies . If this set of states is empty, then Definitely( ) holds and if this set contains only
the final global state then Definitely( ) holds. Note that, unlike detecting Possibly( ), not
all global states need be examined. For example, in Figure 16, suppose that when the global
states of level 2 were constructed, it was determined that 02 satisfied . When constructing
the states of level 3, global state 03 need not be included since it is reachable only through
02
.
The two detection algorithms are linear in the number of global states, but unfor-
tunately the number of global states is where is the maximum number events a
monitored process has executed. There are techniques that can be used to limit the number
of constructed global states. For example, a process need only send a message to 0 when
potentially changes or when learns that has potentially changed . Another tech-
nique is for to send an empty message to all other processes when potentially changes .
These, and other techniques for limiting the number of global states are discussed in [22]. An
alternative approach is to restrict the global predicate to one that can be efficiently detected,
such as the conjunction and disjunction of local predicates [10].

UBLCS-93-1 36
 15 Multiple Monitors

15 Multiple Monitors
There are several good reasons for having multiple monitors observe the same computa-
tion for the purposes of evaluating the same predicate [14]. One such reason is increased
performance—in a large system, interested parties may have the result of the predicate sooner
by asking the monitor that is closest to them.21 Another reason is increased reliability—if
the predicate encodes the condition guarding a critical system action (e.g., shutdown of a
chemical plant), then having multiple monitors will ensure the action despite a bounded
number of failures.
The reactive-architecture solution to GPE based on passive observations can be easily
extended to multiple monitors without modifying its general structure. The only change that
is required is for the processes to use a causal broadcast communication primitive to notify
the group of monitors [12]. In this manner, each monitor will construct a consistent, but not
necessarily the same, observation of the computation. Each observation will correspond to
a (possibly different) path through the global state lattice of the computation. Whether the
results of evaluating predicate by each monitor using local observations are meaningful
depends on the properties of . In particular, if is stable and some monitor observes that
it holds, then all monitors will eventually observe that it holds. For example, in the case of
deadlock detection with multiple monitors, if any one of them detects a deadlock, eventually
all of them will detect the same deadlock since deadlock is a stable property. They may,
however, disagree on the identity of the process that is responsible for creating the deadlock
(the one who issued the last request forming the cycle in the WFG ).
Multiple observations for evaluating nonstable predicates create problems similar to
those discussed in Section 14.2. There are essentially two possibilities for meaningfully
evaluating nonstable predicates over multiple observations. First, the predicate can be
extended using Definitely or Possibly such that it is made independent of the particular
observation but becomes a function of the computation, which is the same for all monitors.
Alternatively, the notification messages can be disseminated to the group of monitors in a
manner such that they all construct the same observation. This can be achieved by using
a causal atomic broadcast primitive that results in a unique total order consistent with causal
precedence for all messages (even those that are concurrent) at all destinations [7,12].
Now consider the case where the monitor is replicated for increased reliability. If
communication or processes in a distributed system are subject to failures, then sending the
same notification message to all monitor replicas using causal delivery is not sufficient for
implementing a causal broadcast primitive. For example if channels are not reliable, some
of the notification messages may be lost such that different monitors effectively observe
different computations. Again, we can accommodate communication and processes failures
in our reactive architecture by using a reliable version of causal broadcast as the communica-
tion primitive [12]. Informally, a reliable causal broadcast, in addition to preserving causal
precedence among message send events, guarantees delivery of a message either to all or
none of the destination processes.22 A formal specification of reliable causal broadcast in the
presence of failures has to be done with care and can be found in [12].
Note that in an asynchronous system subject to failures, reliable causal broadcast is
the best one can hope for in that it is impossible to implement communication primitives

21. In the limit, each process could act as a monitor such that the predicate could be evaluated locally.
22. Note that in [12], this primitive is called casual broadcast without the reliable qualifier since all broadcast
primitives are specified in the presence of failures.

UBLCS-93-1 37
 16 Conclusions

that achieve totally-ordered delivery using deterministic algorithms. Furthermore, in an

environment where processes may fail, and thus certain events never get observed, our
notion of a consistent global state may need to be re-examined. For some global predicates,
the outcome may be sensitive not only to the order in which events are observed, but also to
the order in which failures are observed by the monitors. In such cases, it will be necessary
to extend the causal delivery abstraction to include not only actual messages but also failure
notifications (as is done in systems such as ISIS [2]).

16 Conclusions
We have used the GPE problem as a motivation for studying consistent global states of
distributed systems. Since many distributed systems problems require recognizing certain
global conditions, the construction of consistent global states and the evaluation of predicates
over these states constitute fundamental primitives with which one can build such systems.
We derived two classes of solutions to the GPE problem: a snapshot based one, that
constructs a consistent global state and a reactive-architecture based one, that constucts
a consistent run of the system. To derive these two classes, we have developed a set
of basic concepts and mechanisms for representing and reasoning about computations in
asynchronous distributed systems. These concepts represent generalizations of the notion
of time in order to capture the uncertainty that is inherent in the execution of such systems.
Two kinds of clocks were presented: logical clocks that generalize real-time clocks, and vector
clocks that are incomparable to real-time clocks but exactly capture the causal precedence
relation between events.
We illustrated the practicality of our concepts and mechanisms by applying them to
distributed deadlock detection and distributed debugging. Reactive-architecture solutions
based on passive observations were shown to be more flexible because they capture a run
instead of just a global state. In particular, these solutions can be easily adapted to deal
with nonstable predicates, multiple observations and failures. Each extension can be easily
accommodated by using an appropriate communication primitive for notifications, leaving
the overall reactive architecture unchanged.

Acknowledgments The material on distributed debugging is derived from joint work with
Robert Cooper and Gil Neiger. We are grateful to them for consenting to its inclusion here.
The presentation has benefited greatly from extensive comments by Friedemann Mattern,
Michel Raynal and Fred Schneider on earlier drafts.

References
[1] Baruch Awerbuch. Complexity of network synchronization. Journal of the ACM,
32(4):804–823, October 1985.
[2] K. Birman. The process group approach to reliable distributed computing. Technical
Report TR91-1216, Department of Computer Science, Cornell University, January 1993.
To appear in Communications of the ACM.
[3] K. Birman, A. Schiper, and P. Stephenson. Lightweight causal and atomic group multi-
cast. ACM Transactions on Computer Systems, 9(3):272–314, August 1991.

UBLCS-93-1 38
 REFERENCES

[4] Andrew D. Birrel and Bruce J. Nelson. Implementing remote procedure calls. ACM
Transactions on Computer Systems, 2(1):39–59, February 1984.
[5] K. Mani Chandy and Leslie Lamport. Distributed snapshots: Determining global states
of distributed systems. ACM Transactions on Computer Systems, 3(1):63–75, February
1985.
[6] Robert Cooper and Keith Marzullo. Consistent detection of global predicates. In
ACM/ONR Workshop on Parallel and Distributed Debugging, pages 163–173, Santa Cruz,
California, May 1991.
[7] Flaviu Cristian, H. Aghili, H. Ray Strong, and Danny Dolev. Atomic broadcast: From
simple message diffusion to Byzantine agreement. In Proceedings of the International
Symposium on Fault-Tolerant Computing, pages 200–206, Ann Arbor, Michigan, June
1985. A revised version appears as IBM Technical Report RJ5244.
[8] D. Dolev, J.Y. Halpern, and R. Strong. On the possibility and impossibility of achieving
clock synchronization. In Proceedings of the ACM Symposium on the Theory of Computing,
pages 504–511, April 1984.
[9] C. J. Fidge. Timestamps in message-passing systems that preserve the partial ordering.
In Eleventh Australian Computer Science Conference, pages 55–66, University of Queens-
land, February 1988.
[10] V. K. Garg and B. Waldecker. Unstable predicate detection in distributed programs.
Technical Report TR-92-07-82, University of Texas at Austin, March 1992.
[11] V. Gligor and S. Shattuck. On deadlock detection in distributed systems. IEEE Transac-
tions on Software Engineering, SE-6:435–440, September 1980.
[12] Vassos Hadzilacos and Sam Toueg. Reliable broadcast and agreement algorithms. In
S.J. Mullender, editor, Distributed Systems, chapter 5. ACM Press, 1993.
[13] David Harel and Amir Pnueli. On the development of reactive systems. In K. R. Apt,
editor, Logics and Models of Concurrent Systems, NATO ASI, pages 477–498. Springer-
Verlag, 1985.
[14] J. Helary, C. Jard, N. Plouzeau, and M. Raynal. Detection of stable properties in dis-
tributed applications. In Proceedings of the ACM Symposium on Principles of Distributed
Computing, pages 125–136, Vancouver, British Columbia, August 1987.
[15] M. Frans Kaashoek and Andrew S. Tanenbaum. Group communication in the amoeba
distributed operating system. In Proceedings of the Eleventh International Conference on
Distributed Computer Systems, pages 222–230, Arlington, Texas, May 1991. IEEE Com-
puter Society.
[16] Hermann Kopetz. Sparse time versus dense time in distributed real-time systems. In
Proceedings of the Twelfth International Conference on Distributed Computing Systems, pages
460–467, Yokohama, Japan, June 1992. IEEE Computer Society.
[17] Leslie Lamport. Time, clocks, and the ordering of events in a distributed system.
Communications of the ACM, 21(7):558–565, July 1978.
[18] Leslie Lamport and P. M. Melliar-Smith. Synchronizing clocks in the presence of faults.
Journal of the ACM, 32(1):52–78, January 1985.
[19] Butler Lampson. Atomic transactions. In B. Lampson et. al., editor, Distributed Systems
– Architecture and Implementation: An Advanced Course, volume 105 of Lecture Notes on
Computer Science, pages 246–265. Springer-Verlag, 1981.
[20] H. M. Levy and E. D. Tempero. Modules, objects, and distributed programming: Issues
in rpc and remote object invocation. Software - Practice and Experience, 21(1):77–90,
January 1991.

UBLCS-93-1 39
 REFERENCES

[21] Kai Li and Paul Hudak. Memory coherence in shared virtual memory systems. ACM
Transactions on Computer Systems, 7(4):321–359, November 1989.
[22] Keith Marzullo and Gil Neiger. Detection of global state predicates. In Proceedings of
the Fifth International Workshop on Distributed Algorithms (WDAG-91), Lecture Notes on
Computer Science. Springer-Verlag, Delphi, Greece, October 1991.
[23] Friedemann Mattern. Virtual time and global states of distributed systems. In
Michel Cosnard et. al., editor, Proceedings of the International Workshop on Parallel and
Distributed Algorithms, pages 215–226. North-Holland, October 1989.
[24] Friedemann Mattern. Efficient algorithms for distributed snapshots and global virtual
time approximation. Journal of Parallel and Distributed Computing, 1993. To appear.
[25] J. Misra. Distributed-discrete event simulation. ACM Computing Surveys, 18(1):39–65,
March 1986.
[26] Carroll Morgan. Global and logical time in distributed algorithms. Information Processing
Letters, 20:189–194, May 1985.
[27] Gil Neiger and Sam Toueg. Substituting for real time and common knowledge in
asynchronous distributed systems. In Proceedings of the ACM Symposium on Principles of
Distributed Computing, pages 281–293, Vancouver, British Columbia, August 1987.
[28] Larry L. Peterson, Nick C. Bucholz, and Richard D. Schlichting. Preserving and using
context information in interprocess communication. ACM Transactions on Computer
Systems, 7(3):217–246, August 1989.
[29] M. Raynal. About logical clocks for distributed systems. Operating Systems Review,
26(1):41–48, January 1992.
[30] M. Raynal, A. Schiper, and S. Toueg. The causal ordering abstraction and a simple way
to implement it. Information Processing Letters, 39(6):343–350, September 1991.
[31] A. Sandoz and A. Schiper. A characterization of consistent distributed snapshots using
causal order. Technical Report 92-14, Departement d’Informatique, Ecole Polytechnique
Federale de Lausanne, Switzerland, October 1992.
[32] A. Schiper, J. Eggli, and A. Sandoz. A new algorithm to implement causal ordering. In
J.-C. Bermond and M. Raynal, editors, Proceedings of the Third International Workshop on
Distributed Algorithms, volume 392 of Lecture Notes on Computer Science, pages 219–232,
Nice, France, September 1989. Springer-Verlag.
[33] Fred Schneider. Replication management using the state machine approach. In S.J.
Mullender, editor, Distributed Systems, chapter 7. ACM Press, 1993.
[34] Reinhard Schwarz and Friedemann Mattern. Detecting causal relationships in dis-
tributed computations: In search of the Holy Grail. Technical Report SFB124-15/92,
Department of Computer Science, University of Kaiserslautern, Kaiserslautern, Ger-
many, December 1992.
[35] Kim Taylor. The role of inhibition in asynchronous consistent-cut protocols. In J.-
C. Bermond and M. Raynal, editors, Proceedings of the Third International Workshop on
Distributed Algorithms, volume 392 of Lecture Notes on Computer Science, pages 280–291.
Springer-Verlag, Nice, France, September 1989.

UBLCS-93-1 40