Berkeley latency-mar2012

Achieving Rapid Response Times in
Large Online Services
Jeff Dean
Google Fellow
jeff@google.com

Monday, March 26, 2012

Faster Is Better


Large Fanout Services
query

Ad System Frontend Web Server

Super root Cache servers

News
Local
Video
Images Blogs
Books

Web


Why Does Fanout Make Things Harder?
• Overall latency ≥ latency of slowest component
– small blips on individual machines cause delays
– touching more machines increases likelihood of delays

• Server with 1 ms avg. but 1 sec 99%ile latency
– touch 1 of these: 1% of requests take ≥1 sec
– touch 100 of these: 63% of requests take ≥1 sec


One Approach: Squash All Variability
• Careful engineering all components of system

• Possible at small scale
– dedicated resources
– complete control over whole system
– careful understanding of all background activities
– less likely to have hardware fail in bizarre ways

• System changes are difficult
– software or hardware changes affect delicate balance

Not tenable at large scale: need to share resources


Shared Environment
• Huge benefit: greatly increased utilization

• ... but hard to predict effects increase variability
– network congestion
– background activities
– bursts of foreground activity
– not just your jobs, but everyone else’s jobs, too

• Exacerbated by large fanout systems


Shared Environment

Linux


Shared Environment

file system
chunkserver
Linux


Shared Environment

file system scheduling
chunkserver system
Linux


Shared Environment

various other
system services
chunkserver system
Linux


Shared Environment

Bigtable
tablet server

various other
system services
chunkserver system
Linux


Shared Environment

cpu intensive
job
Bigtable
tablet server

various other
system services
chunkserver system
Linux


Shared Environment

cpu intensive
job
random Bigtable
MapReduce #1 tablet server

various other
system services
chunkserver system
Linux


Shared Environment

random
app #2
cpu intensive random
job app
random Bigtable
MapReduce #1 tablet server

various other
system services
chunkserver system
Linux


Basic Latency Reduction Techniques
• Differentiated service classes
– prioritized request queues in servers
– prioritized network traffic

• Reduce head-of-line blocking
– break large requests into sequence of small requests

• Manage expensive background activities
– e.g. log compaction in distributed storage systems
– rate limit activity
– defer expensive activity until load is lower


Synchronized Disruption
• Large systems often have background daemons
– various monitoring and system maintenance tasks

• Initial intuition: randomize when each machine
performs these tasks
– actually a very bad idea for high fanout services
• at any given moment, at least one or a few machines are slow

• Better to actually synchronize the disruptions
– run every five minutes “on the dot”
– one synchronized blip better than unsynchronized


Tolerating Faults vs. Tolerating Variability
• Tolerating faults:
– rely on extra resources
• RAIDed disks, ECC memory, dist. system components, etc.
– make a reliable whole out of unreliable parts

• Tolerating variability:
– use these same extra resources
– make a predictable whole out of unpredictable parts

• Times scales are very different:
– variability: 1000s of disruptions/sec, scale of milliseconds
– faults: 10s of failures per day, scale of tens of seconds


Latency Tolerating Techniques
• Cross request adaptation
– examine recent behavior
– take action to improve latency of future requests
– typically relate to balancing load across set of servers
– time scale: 10s of seconds to minutes

• Within request adaptation
– cope with slow subsystems in context of higher level
request
– time scale: right now, while user is waiting


Fine-Grained Dynamic Partitioning
• Partition large datasets/computations
– more than 1 partition per machine (often 10-100/machine)
– e.g. BigTable, query serving systems, GFS, ...

Master

17 9

1 3 2 8 4 12 7
... ... ... ...
Machine 1 Machine 2 Machine 3 Machine N


Load Balancing
• Can shed load in few percent increments
– prioritize shifting load when imbalance is more severe

Master

17 9

1 3 2 8 4 12 7
... ... ... ...


Load Balancing

Master

17 9

1 3 2 8 4 12 7
... ... ... ...
Overloaded!


Load Balancing

Master

17

1 3 2 9 8 4 12 7
... ... ... ...
Overloaded!


Load Balancing

Master

17

1 3 2 9 8 4 12 7
... ... ... ...


Speeds Failure Recovery
• Many machines each recover one or a few partition
– e.g. BigTable tablets, GFS chunks, query serving shards

Master

17

1 3 2 9 8 4 12 7
... ... ... ...



Master

2 9 8 4 12 7
... ... ...



Master

1 3 17

2 9 8 4 12 7
... ... ...


Selective Replication
• Find heavily used items and make more replicas
– can be static or dynamic

• Example: Query serving system
– static: more replicas of important docs
– dynamic: more replicas of Chinese documents as
Chinese query load increases
Master

... ... ... ...


Latency-Induced Probation
• Servers sometimes become slow to respond
– could be data dependent, but...
– often due to interference effects
• e.g. CPU or network spike for other jobs running on shared server

• Non-intuitive: remove capacity under load to
improve latency (?!)
• Initiate corrective action
– e.g. make copies of partitions on other servers
– continue sending shadow stream of requests to server
• keep measuring latency
• return to service when latency back down for long enough


Handling Within-Request Variability
• Take action within single high-level request

• Goals:
– reduce overall latency
– don’t increase resource use too much
– keep serving systems safe


Data Independent Failures

query


Canary Requests (2)

query


Backup Requests
req 3 req 5 req 8
req 6

Replica 1 Replica 2 Replica 3

Client


Backup Requests
req 3 req 5 req 8
req 6


Client9
req


Backup Requests
req 3 req 5 req 8
req 6
req 9


Client


Backup Requests
req 3 req 5 req 8
req 6 req 9
req 9


Client


Backup Requests
req 3 req 8
req 6 req 9
req 9


Client


Backup Requests
req 3 req 8
req 6
req 9

Replica 1 reply
Replica 2 Replica 3

Client


Backup Requests
req 3 req 8
req 6
req 9


reply
Client


Backup Requests
req 3 req 8
req 6
req 9


“Cancel req 9”

reply
Client


Backup Requests
req 3 req 8
req 6
req 9

“Cancel req 9”


reply
Client


Backup Requests
req 3 req 8
req 6


reply
Client


Backup Requests Effects

• In-memory BigTable lookups
– data replicated in two in-memory tables
– issue requests for 1000 keys spread across 100 tablets
– measure elapsed time until data for last key arrives




Avg Std Dev 95%ile 99%ile 99.9%ile
No backups 33 ms 1524 ms 24 ms 52 ms 994 ms
Backup after 10 ms 14 ms 4 ms 20 ms 23 ms 50 ms




Avg Std Dev 95%ile 99%ile 99.9%ile
No backups 33 ms 1524 ms 24 ms 52 ms 994 ms

• Modest increase in request load:
– 10 ms delay: <5% extra requests; 50 ms delay: <1%


Backup Requests w/ Cross-Server Cancellation

req 3 req 5
req 6

Server 1 Server 2

Client



req 3 req 5
req 6

Server 1 Server 2

req 9
Client



req 3 req 5
req 6

req 9
also: server 2

Server 1 Server 2

req 9
Client

Each request identifies other server(s) to which request might be sent



req 3 req 5
req 6 req 9
also: server 1
req 9
also: server 2

Server 1 Server 2

Client




req 3
req 6 req 9
also: server 1
req 9
also: server 2

Server 1 Server 2

Client




req 3
req 6 req 9
also: server 1
req 9
also: server 2

Server 1 Server 2
“Server 2: Starting req 9”

Client




req 3
req 6 req 9
also: server 1

Server 1 Server 2

Client




req 3
req 6 req 9
also: server 1

Server 1 Server 2
reply

Client




req 3
req 6 req 9
also: server 1

Server 1 Server 2

Client
reply



Backup Requests: Bad Case

req 3 req 5

Server 1 Server 2

Client



req 3 req 5

Server 1 Server 2

req 9
Client



req 3 req 5

req 9
also: server 2

Server 1 Server 2

req 9
Client



req 3 req 5

req 9 req 9
also: server 2 also: server 1

Server 1 Server 2

Client



req 9 req 9

Server 1 Server 2

Client



req 9 req 9

Server 1 Server 2
reply

Client



req 9 req 9

Server 1 Server 2

Client
reply



• Read operations in distributed file system client
– send request to first replica
– wait 2 ms, and send to second replica
– servers cancel request on other replica when starting read
• Time for bigtable monitoring ops that touch disk




Cluster state Policy 50%ile 90%ile 99%ile 99.9%ile
Mostly idle No backups 19 ms 38 ms 67 ms 98 ms
Backup after 2 ms 16 ms 28 ms 38 ms 51 ms



• Time for bigtable monitoring ops that touch disk -43%






+Terasort No backups 24 ms 56 ms 108 ms 159 ms



• Time for bigtable monitoring ops that touch disk -38%








Backups cause about ~1% extra disk reads






Backups w/big sort job gives same read latencies as no backups w/ idle cluster!


Backup Request Variants

• Many variants possible:

• Send to third replica after longer delay
– sending to two gives almost all the benefit, however.

• Keep requests in other queues, but reduce priority

• Can handle Reed-Solomon reconstruction similarly


Tainted Partial Results

• Many systems can tolerate inexact results
– information retrieval systems
• search 99.9% of docs in 200ms better than 100% in 1000ms
– complex web pages with many sub-components
• e.g. okay to skip spelling correction service if it is slow

• Design to proactively abandon slow subsystems
– set cutoffs dynamically based on recent measurements
• can tradeoff completeness vs. responsiveness
– important to mark such results as tainted in caches


Hardware Trends

• Some good:
– lower latency networks make things like backup request
cancellations work better

• Some not so good:
– plethora of CPU and device sleep modes save power,
but add latency variability
– higher number of “wimpy” cores => higher fanout =>
more variability

• Software techniques can reduce variability despite
increasing variability in underlying hardware


Conclusions

• Tolerating variability
– important for large-scale online services
– large fanout magnifies importance
– makes services more responsive
– saves significant computing resources

• Collection of techniques
– general good engineering practices
• prioritized server queues, careful management of background activities
– cross-request adaptation
• load balancing, micro-partitioning
– within-request adaptation
• backup requests, backup requests w/ cancellation, tainted results


Thanks

• Joint work with Luiz Barroso and many others at
Google

• Questions?


Berkeley latency-mar2012

More Related Content

What's hot (13)

Viewers also liked (8)

Similar to Berkeley latency-mar2012 (20)

More from drewz lin (20)

Recently uploaded (20)

Berkeley latency-mar2012