Lecture 8

Virtual Circuits, ATM, MPLS

David Andersen
School of Computer Science
Carnegie Mellon University

15-441 Networking, Fall 2006

http://www.cs.cmu.edu/~srini/15-441/F06/

Outline

 Exam discussion
 Layering review (bridges, routers, etc.)
» Exam section C.
 Circuit switching refresher
 Virtual Circuits - general
» Why virtual circuits?
» How virtual circuits? -- tag switching!
 Two modern implementations
» ATM - teleco-style virtual circuits
» MPLS - IP-style virtual circuits

Page ‹#›
 Exam stats

Max/avg/min: 90 / 63 / 20
A B C D
19.6 17.9 12.8 11.0
57.6% 74.8% 58.3% 68.6%

Common Exam Problems

 Routing: No one big problem; many small

misunderstandings. Please check your
scores.
 Short answer: Many incorrect round-trip
times vs. one-way times.
 DNS
» Always sends the full query! (e.g.
“ra1.streaming.npr.org”, not just “npr.org”)
» Clients don’t recurse; the local recursive DNS server
does. Could run on clients, but usually doesn’t.
 Routing and bridging and addressing…

Page ‹#›
 Packet Switching

 Source sends information as self-contained

packets that have an address.
» Source may have to break up single message in multiple
 Each packet travels independently to the
destination host.
» Routers and switches use the address in the packet to
determine how to forward the packets
 Destination recreates the message.
 Analogy: a letter in surface mail.

Circuit Switching

 Source first establishes a connection (circuit)

to the destination.
» Each router or switch along the way may reserve some
bandwidth for the data flow
 Source sends the data over the circuit.
» No need to include the destination address with the data
since the routers know the path
 The connection is torn down.
 Example: telephone network.

Page ‹#›
 Circuit Switching
Discussion

 Traditional circuits: on each hop, the circuit

has a dedicated wire or slice of bandwidth.
» Physical connection - clearly no need to include
addresses with the data
 Advantages, relative to packet switching:
» Implies guaranteed bandwidth, predictable performance
» Simple switch design: only remembers connection
information, no longest-prefix destination address look
up
 Disadvantages:
» Inefficient for bursty traffic (wastes bandwidth)
» Delay associated with establishing a circuit
 Can we get the advantages without (all) the
disadvantages?

Virtual Circuits

 Each wire carries many “virtual” circuits.

» Forwarding based on virtual circuit (VC) identifier
– IP header: src, dst, etc.
– Virtual circuit header: just “VC”
» A path through the network is determined for each VC when the
VC is established
» Use statistical multiplexing for efficiency
 Can support wide range of quality of service.
» No guarantees: best effort service
» Weak guarantees: delay < 300 msec, …
» Strong guarantees: e.g. equivalent of physical circuit

Page ‹#›
 Packet Switching and
Virtual Circuits: Similarities
 “Store and forward” communication based on an
address.
» Address is either the destination address or a VC identifier
 Must have buffer space to temporarily store packets.
» E.g. multiple packets for some destination arrive simultaneously
 Multiplexing on a link is similar to time sharing.
» No reservations: multiplexing is statistical, i.e. packets are
interleaved without a fixed pattern
» Reservations: some flows are guaranteed to get a certain
number of “slots”

D B C B A A

Virtual Circuits Versus

Packet Switching

 Circuit switching:
» Uses short connection identifiers to forward packets
» Switches know about the connections so they can more
easily implement features such as quality of service
» Virtual circuits form basis for traffic engineering: VC
identifies long-lived stream of data that can be scheduled
 Packet switching:
» Use full destination addresses for forwarding packets
» Can send data right away: no need to establish a
connection first
» Switches are stateless: easier to recover from failures
» Adding QoS is hard
» Traffic engineering is hard: too many packets!

10

Page ‹#›
 Circuit Switching

Switch

Input
Output
Ports
Ports

Connects (electrons or bits) ports to ports

11

Packet switched vs. VC

Payload VCI Payload Dst

1 3
R1 packet
A 2 R2 forwarding
1 3 4 1 3
table:
R1 R4 Dst
2 4 2 4 Dst R2
1 3
B R3
2 4
Different paths to
R1 VC table: same destination!

VC 1 R2 (useful for traffic

engineering!)
VC 2 R3 12

Page ‹#›
 Virtual Circuit

Payload VCI Payload

1 3
A 2 R2
1 3 4 1 3
R1 R4 Dst
2 4 2 4
1 3
B R3
2 4 Challenges:

R1 VC table: - How to set up path?

R2 VC table:
VC 5 R2 - How to assign IDs??
VC 5 R4

13

Connections and Signaling

 Permanent vs. switched virtual connections (PVCs, SVCs)

» static vs. dynamic. PVCs last “a long time”
– E.g., connect two bank locations with a PVC that looks like a circuit
– SVCs are more like a phone call
» PVCs administratively configured (but not “manually”)
» SVCs dynamically set up on a “per-call” basis
 Topology
» point to point
» point to multipoint
» multipoint to multipoint
 Challenges:
» How to configure these things?
– What VCI to use?
– Setting up the path

14

Page ‹#›
 Virtual Circuit Switching:
Label (“tag”) Swapping
1 3
A 2 R2
1 3 4 1 3
R1 R4 Dst
2 4 2 4
1 3
B R3
2 4

 Global VC ID allocation -- ICK! Solution: Per-link uniqueness.

Change VCI each hop.

Input Port Input VCI Output Port Output VCI

R1: 1 5 3 9

R2: 2 9 4 2

R4: 1 2 3 5 15

Label (“tag”) Swapping

 Result: Signalling protocol must only find

per-link unused VCIs.
» “Link-local scope”
» Connection setup can proceed hop-by-hop.
– Good news for our setup protocols!

16

Page ‹#›
 PVC connection setup

 Manual?
» Configure each switch by hand. Ugh.
 Dedicated signalling protocol
» E.g., what ATM uses
 Piggyback on routing protocols
» Used in MPLS. E.g., use BGP to set up

17

SVC Connection Setup

calling network called
party party

SETUP

SETUP

CONNECT

CONNECT

CONNECT
ACK
CONNECT
ACK

18

Page ‹#›
 Virtual Circuits In Practice

 ATM: Teleco approach

» Kitchen sink. Based on voice, support file transfer, video, etc.,
etc.
» Intended as IP replacement. That didn’t happen. :)
» Today: Underlying network protocol in many teleco networks.
E.g., DSL speaks ATM. IP over ATM in some cases.
 MPLS: The “IP Heads” answer to ATM
» Stole good ideas from ATM
» Integrates well with IP
» Today: Used inside some networks to provide VPN support,
traffic engineering, simplify core.
 Other nets just run IP.
 Older tech: Frame Relay
» Only provided PVCs. Used for quasi-dedicated 56k/T1 links
between offices, etc. Slower, less flexible than ATM.

19

Asynchronous Transfer Mode:

ATM

 Connection-oriented, packet-switched
» (e.g., virtual circuits).
 Teleco-driven. Goals:
» Handle voice, data, multimedia
» Support both PVCs and SVCs
» Replace IP. (didn’t happen…)
 Important feature: Cell switching

20

Page ‹#›
 Cell Switching

 Small, fixed-size cells

[Fixed-length data][header]

 Why?
» Efficiency: All packets the same
– Easier hardware parallelism, implementation
» Switching efficiency:
– Lookups are easy -- table index.
» Result: Very high cell switching rates.
» Initial ATM was 155Mbit/s. Ethernet was 10Mbit/s at the same
time. (!)
 How do you pick the cell size?

21

ATM Features

 Fixed size cells (53 bytes).

» Why 53?
 Virtual circuit technology using hierarchical virtual
circuits (VP,VC).
 PHY (physical layer) processing delineates cells by
frame structure, cell header error check.
 Support for multiple traffic classes by adaptation layer.
» E.g. voice channels, data traffic
 Elaborate signaling stack.
» Backwards compatible with respect to the telephone standards
 Standards defined by ATM Forum.
» Organization of manufacturers, providers, users

22

Page ‹#›
 Why 53 Bytes?

 Small cells favored by voice applications

» delays of more than about 10 ms require echo
cancellation
» each payload byte consumes 125 µs (8000
samples/sec)
 Large cells favored by data applications
» Five bytes of each cell are overhead
 France favored 32 bytes
» 32 bytes = 4 ms packetization delay.
» France is 3 ms wide.
» Wouldn’t need echo cancellers!
 USA, Australia favored 64 bytes
» 64 bytes = 8 ms
» USA is 16 ms wide
» Needed echo cancellers anyway, wanted less
overhead
 Compromise
23

ATM Adaptation Layers

1 2 3 4 5
synchronous asynchronous
constant variable bit rate
connection-oriented connectionless

 AAL 1: audio, uncompressed video

 AAL 2: compressed video
 AAL 3: long term connections
 AAL 4/5: data traffic
 AAL5 is most relevant to us…

24

Page ‹#›
 AAL5 Adaptation Layer

data pad ctl len CRC

...

ATM payload
header (48 bytes)

includes EOF flag

Pertinent part: Packets are spread across multiple ATM

cells. Each packet is delimited by EOF flag in cell. 25

ATM Packet Shredder Effect

 Cell loss results in packet loss.

» Cell from middle of packet: lost packet
» EOF cell: lost two packets
» Just like consequence of IP fragmentation, but VERY small
fragments!
 Even low cell loss rate can result in high packet loss
rate.
» E.g. 0.2% cell loss -> 2 % packet loss
» Disaster for TCP
 Solution: drop remainder of the packet, i.e. until EOF
cell.
» Helps a lot: dropping useless cells reduces bandwidth and
lowers the chance of later cell drops
» Slight violation of layers
» Discovered after early deployment experience with IP over ATM.

26

Page ‹#›
 IP over ATM

 When sending IP packets over an ATM

network, set up a VC to destination.
» ATM network can be end to end, or just a partial path
» ATM is just another link layer
 Virtual connections can be cached.
» After a packet has been sent, the VC is maintained so that
later packets can be forwarded immediately
» VCs eventually times out
 Properties.
– Overhead of setting up VCs (delay for first packet)
– Complexity of managing a pool of VCs
+ Flexible bandwidth management
+ Can use ATM QoS support for individual connections
(with appropriate signaling support)
27

IP over ATM
Static VCs

 Establish a set of “ATM

pipes” that defines
connectivity between routers.
 Routers simply forward
packets through the pipes.
» Each statically configured VC
looks like a link
 Properties.
– Some ATM benefits are lost (per
flow QoS)
+ Flexible but static bandwidth
management
+ No set up overheads

28

Page ‹#›
 ATM Discussion
 At one point, ATM was viewed as a replacement for IP.
» Could carry both traditional telephone traffic (CBR circuits)
and other traffic (data, VBR)
» Better than IP, since it supports QoS
 Complex technology.
» Switching core is fairly simple, but
» Support for different traffic classes
» Signaling software is very complex
» Technology did not match people’s experience with IP
– deploying ATM in LAN is complex (e.g. broadcast)
– supporting connection-less service model on
connection-based technology
» With IP over ATM, a lot of functionality is replicated
 Currently used as a datalink layer supporting IP.

29

Multi Protocol Label Switching -

MPLS
 Selective combination of VCs + IP
» Today: MPLS useful for traffic engineering, reducing core
complexity, and VPNs
 Core idea: Layer 2 carries VC label
» Could be ATM (which has its own tag)
» Could be a “shim” on top of Ethernet/etc.:
» Existing routers could act as MPLS switches just by examining
that shim -- no radical re-design. Gets flexibility benefits, though
not cell switching advantages

Layer 3 (IP) header

Layer 3 (IP) header
MPLS label
Layer 2 header
Layer 2 header
30

Page ‹#›
 MPLS + IP

 Map packet onto Forward Equivalence Class (FEC)

» Simple case: longest prefix match of destination address
» More complex if QoS of policy routing is used
 In MPLS, a label is associated with the packet when it
enters the network and forwarding is based on the
label in the network core.
» Label is swapped (as ATM VCIs)
 Potential advantages.
» Packet forwarding can be faster
» Routing can be based on ingress router and port
» Can use more complex routing decisions
» Can force packets to followed a pinned route

31

MPLS core, IP interface

MPLS tag MPLS tag

assigned stripped
IP
IP
IP IP
1 3 C
A 2 R2
1 3 4 1 3
R1 R4
2 4 2 4
1 3
B R3 D
2 4
MPLS forwarding in core
32

Page ‹#›
 MPLS use case #1: VPNs

10.1.0.0/24
10.1.0.0/24
1 3 C
A 2 R2
1 3 4 1 3
R1 R4
2 4 2 4
1 3
B R3 D
2 4
10.1.0.0/24 10.1.0.0/24

MPLS tags can differentiate green VPN from orange VPN.

33

MPLS use case #2: Reduced

State Core
EBGP EBGP C
A R2
A-> C pkt R1 R4
IP Core
Internal routers must
R3
know all C destinations

1 3 EBGP C
A 2 R2
1 3 4 1 3
R1 MPLS Core R4
2 4 2 4
R1 uses MPLS tunnel to R4. 1
R3
3
2 4
. R4 know routes, but
R1 and
R2 and R3 don’t. 34

Page ‹#›
 MPLS use case #3: Traffic
Engineering

 As discussed earlier -- can pick routes based

upon more than just destination
 Used in practice by many ISPs, though
certainly not all.

35

MPLS Mechanisms

 MPLS packet forwarding: implementation of

the label is technology specific.
» Could be ATM VCI or a short extra “MPLS” header

 Supports stacked labels.

» Operations can be “swap” (normal label swapping),
“push” and “pop” labels.
– VERY flexible! Like creating tunnels, but much
simpler -- only adds a small label.

Label CoS S TTL

20 3 1 8
36

Page ‹#›
 MPLS Discussion

 Original motivation.
» Fast packet forwarding:
– Use of ATM hardware
– Avoid complex “longest prefix” route lookup
– Limitations of routing table sizes
» Quality of service
 Currently mostly used for traffic engineering
and network management.
» LSPs can be thought of as “programmable links” that can
be set up under software control
» on top of a simple, static hardware infrastructure

37

Take Home Points

 Costs/benefits/goals of virtual circuits

 Cell switching (ATM)
» Fixed-size pkts: Fast hardware
» Packet size picked for low voice jitter. Understand trade-
offs.
» Beware packet shredder effect (drop entire pkt)
 Tag/label swapping
» Basis for most VCs.
» Makes label assignment link-local. Understand
mechanism.
 MPLS - IP meets virtual circuits
» MPLS tunnels used for VPNs, traffic engineering, reduced
core routing table sizes

38

Page ‹#›
 --- Extra Slides ---

Extra information if you’re curious.

39

ATM Traffic Classes

 Constant Bit Rate (CBR) and Variable Bit Rate

(VBR).
» Guaranteed traffic classes for different traffic types.
 Unspecified Bit Rate (UBR).
» Pure best effort with no help from the network
 Available Bit Rate (ABR).
» Best effort, but network provides support for congestion
control and fairness
» Congestion control is based on explicit congestion
notification
– Binary or multi-valued feedback
» Fairness is based on Max-Min Fair Sharing.
(small demands are satisfied, unsatisfied demands share equally)

40

Page ‹#›
 LAN Emulation

 Motivation: making a non-broadcast

technology work as a LAN.
» Focus on 802.x environments
 Approach: reuse the existing interfaces, but
adapt implementation to ATM.
» MAC - ATM mapping
» multicast and broadcast
» bridging
» ARP
 Example: Address Resolution “Protocol”
uses an ARP server instead of relying on
broadcast.
41

Further reading - MPLS

 MPLS isn’t in the book - sorry. Juniper has a

few good presentations at NANOG (the North
American Network Operators Group; a big
collection of ISPs):
» http://www.nanog.org/mtg-0310/minei.html
» http://www.nanog.org/mtg-0402/minei.html
» Practical and realistic view of what people
are doing _today_ with MPLS.

42

Page ‹#›
 IP Switching
 How to use ATM hardware without the software.
» ATM switches are very fast data switches
» software adds overhead, cost
 The idea is to identify flows at the IP level and to create
specific VCs to support these flows.
» flows are identified on the fly by monitoring traffic
» flow classification can use addresses, protocol types, ...
» can distinguish based on destination, protocol, QoS
 Once established, data belonging to the flow bypasses
level 3 routing.
» never leaves the ATM switch
 Interoperates fine with “regular” IP routers.
» detects and collaborates with neighboring IP switches

43

IP Switching Example

IP IP IP

ATM ATM ATM

44

Page ‹#›
 IP Switching Example

IP IP IP

ATM ATM ATM

45

IP Switching Example

IP IP IP

ATM ATM ATM

46

Page ‹#›
 Another View

IP IP IP IP
ATM ATM

IP
IP
ATM
ATM
IP IP
ATM ATM IP
IP
ATM
ATM

IP IP

47

IP Switching
Discussion

 IP switching selectively optimizes the

forwarding of specific flows.
» Offloads work from the IP router, so for a given size
router, a less powerful forwarding engine can be used
» Can fall back on traditional IP forwarding if there are
failures
 IP switching couples a router with an ATM
switching using the GSMP protocol.
» General Switch Management Protocol
 IP switching can be used for flows with
different granularity.
» Flows belonging to an application .. Organization
» Controlled by the classifier

48

Page ‹#›
 An Alternative
Tag Switching

 Instead of monitoring traffic to identify flows to

optimize, use routing information to guide the creation
of “switched” paths.
» Switched paths are set up as a side effect of filling in forwarding
tables
 Generalize to other types of hardware.
 Also introduced stackable tags.
» Made it possible to temporarily merge flows and to demultiplex
them without doing an IP route lookup
» Requires variable size field for tag

AC
A A

B B
BC
49

IP Switching
versus Tag Switching

 Flows versus routes.

» tags explicitly cover groups of routes
» tag bindings set up as part of route establishment
» flows in IP switching are driven by traffic and detected by
“filters”
– Supports both fine grain application flows and
coarser grain flow groups
 Stackable tags.
» provides more flexibility
 Generality
» IP switching focuses on ATM
» not clear that this is a fundamental difference

50

Page ‹#›
 Packets over SONET

 Same as statically
configured ATM
pipes, but pipes are mux
SONET channels. OC-48
 Properties.
– Bandwidth
management is mux
much less flexible
+ Much lower
transmission
overhead (no ATM
headers)
mux

51

Page ‹#›