How to scale a distributed system

Henry Robinson
@henryr / [email protected]
What is this, and who’s it for?
§ Lessons learned from the trenches building distributed systems for 8+ years at Cloudera and in
open source communities.
What is this, and who’s it for?
§ Lessons learned from the trenches building distributed systems for 8+ years at Cloudera and in
open source communities.




§ Not:
§ A complete course in distributed systems theory (but boy do I have references for you)
§ Always specific to distributed systems
§ Complete
§ Signed off by experts
§ A panacea (sorry)
…and you are?
§ Distributed systems dilettante


§ Some years in graduate school for distributed systems




..followed by some years in industry for the same thing.


§ Some writing on my blog: http://the-paper-trail.org/


§ A community: https://dist-sys-slack.herokuapp.com/ for the invite

Today
§ Primitives


§ Practices


§ Possibility


§ Papers
Today
§ Primitives - what are the concepts, and nouns, that it’s important to know?


§ Practices - what are good habits in distributed systems design?


§ Possibility - how should we think - if at all - about formal impossibility?


§ Papers - you don’t have time to read everything? Join the club.
[spoiler: everyone argues

about CAP, forever]
1. Primitives
Basic concepts
§ Processes may fail.

§ There is no particularly good way to tell that they have done so.

§ Almost always better to err on the side of caution.

Basic concepts
1. Failure detectors
2. Symmetry breaking (with leader election as an example)
3. Fault models
4. Replicated state machines
5. Quorums
6. Logical time
7. Coordination: broadcast, consensus, commit protocols
2. Practices
Always Be sCaling
What do we talk about, when we talk about scaling?
§ Scaling (up) means more. Of everything.

§“what happens to the behavioral characteristics of my system as

the operational parameters increase?”

§ Not just number of nodes.

Why are we scaling? Not just increased load.
§ Commodity hardware revolution made incremental capacity improvements possible.

§ The operational mode of the software we build has changed: availability is the sword by which web
properties live or die.

§ Redundancy is the basic conceptual approach to providing availability

§ Adding more processing power is how we provide redundancy; i.e. we scale our systems up.
Scalability axes

§ One rarely considered scalability axis:

more failures. (and more types of
failure)
Scalability axes

§ GFS Paper (SOSP 2003)

§ One rarely considered scalability axis:

more failures. (and more types of
failure)
Scalability axes

§ GFS Paper (SOSP 2003)

§ One rarely considered scalability axis:

more failures. (and more types of
failure)
Apache Impala has to scale with respect to…
Apache Impala has to scale with respect to…
§ Query complexity


§ Queries per second


§ Cluster size


§ Node CPU / memory


§ Degree of per-node parallelism


§ Number of clients per node


§ Number of clients per cluster


§ Number of tables


§ Number of partitions per table

Apache Impala has to scale with respect to…
§ Query complexity

§ Number of columns per table

§ Queries per second

§ Data size per table

§ Cluster size

§ Intermediate result size 

§ Node CPU / memory

§ Kerberos ticket grants
§ Degree of per-node parallelism


§ Number of clients per node


§ Number of clients per cluster


§ Number of tables


§ Number of partitions per table

Scale is a fundamental design consideration

Just like security, include scalability in your thinking from day one.

Scalability behaviors are usually discontinuous - they exhibit phase changes rather than gradual
improvement. (20->50 nodes, not 20->22)

That means you can clearly identify scaling boundaries. Do this wherever possible. The rest of the
your team - and the systems you interact with - will thank you for it.

It also means that, by attacking the scaling boundary, you can have a large impact - when the time is
right.
Draw your borders before you drive off a cliff
Draw your borders before you drive off a cliff

super-linear costs will

eventually dominate
Decompose system properties into
safety and liveness
System invariants
Safety Liveness
System invariants
Safety Liveness

“Nothing bad ever happens!”

For example:

§ Queries never return incorrect results


§ Corrupt data is never written to disk


§ Data is never read remotely


§ Only one leader exists at any time

System invariants
Safety Liveness

“Nothing bad ever happens!” “Something good eventually happens!”

For example: For example:

§ Queries never return incorrect results
 § New nodes eventually join the cluster


§ Corrupt data is never written to disk
 § All queries complete


§ Data is never read remotely
 § Some data gets written to disk on INSERT

§ Only one leader exists at any time

System invariants
Safety Liveness

“Nothing bad ever happens!” “Something good eventually happens!”

For example: For example:

§ Queries never return incorrect results
 § New nodes eventually join the cluster


All
§ Corrupt data is system
never properties
written to disk
 § Allcan be
queries described
complete
 as
a combination
§ Data is never read remotely

of safety and liveness properties.
§ Some data gets written to disk on INSERT

§ Only one leader exists at any time

Example: Impala’s query liveness and safety
§ For queries, liveness means “all queries eventually complete”

(note I didn’t say they complete successfully)


§ Safety property is more interesting. Choice between:

1. Query never returns anything but its full result set
2. Query must return anything, but must signal an error when it does.

§ Impala chose option #2, despite #1 being much more attractive.

§ Why?
Example: Impala’s query liveness and safety
§ It’s obviously better to always return complete results, but failures make that extremely hard.

§ If Impala had tried to enforce strong query safety from day 1, it would never have been a success:
achieving performance goals would have been much harder.

§ Instead, make fault tolerance trivial by weakening the definition. By definition, such a system scales
better.
Think global, 

act local.
Coordination costs
§ Coordination: getting different processes to agree on some shared fact.

§ Coordination is incredibly costly in distributed systems and the cost increases with the number of
participants.

§ This is the reason most ZooKeeper deployments are 3-5 nodes.

Avoid coordination wherever possible
§ Mostly got this right in Impala:

§ Metadata consistent on session level (sticky to one machine) -> no coordination required

§ Data processing is heavily parallel.

§ Coordination happens almost entirely at distinguished coordinator node, asynchronously wrt to

query execution
Example: synchronous DDL

§ Some users wanted cross-session metadata consistency, i.e. I create a table, you can instantly see
it.

§ Problem: symmetry of Impala’s architecture means every Impala daemon needs to see all updates
synchronously.

§ Latency of these operations is by definition pessimal.

Small control plane,
big data plane
Two types of communication

§ Communication in distributed systems serves roughly one of two purposes:

§ Control logic tells processes what to do next


§ Data flow exchanges data between processes for computation


Data vs control
Data protocols

§ Simple protocols

§ Typically need local-state only


§ Very high data volume


§ Heavy resource consumption


§ Highly scalable


§ Dominates CPU execution time


Data vs control
Data protocols Control protocols

§ Simple protocols § Complex protocols

§ Typically need local-state only
 § Global view of cluster state

§ Very high data volume
 § Relatively small data volume

§ Heavy resource consumption
 § Lightweight resource consumption

§ Highly scalable
 § Not highly scalable

§ Dominates CPU execution time
 § Low relative cost

3. (im)possibility
YOU CAN’T DO THAT!
§ Nothing trips up Distributed Systems Twitter faster than impossibility results


§ Two camps:
§ “your system doesn’t beat CAP, so I don’t care”
§ “I don’t care about CAP, it’s really unlikely I’ll lose that transaction”

§ Impossibility results - and there are a lot of them - tell us about some fundamental tension. But they
are completely silent on practicalities. Just because you can’t do something, doesn’t mean you
shouldn’t try.


§ The best way to think about impossibility is to recognize the safety and liveness tension that a
result represents. 


§ Decide which you’re willing to give up. 


§ And then protect the other at all cost. 


4. Papers
Read papers.
Read papers.
Not too many.
 Read papers.
Not too many.
Mostly real systems papers.
Thank you!