Message Queues at salesforce.com

The subtitle goes here

Vijay Devadhar

Developer

Salesforce.com

Safe Harbor

Safe harbor statement under the Private Securities Litigation Reform Act of 1995:

This presentation may contain forward-looking statements that involve risks, uncertainties, and assumptions. If any such uncertainties materialize or if

any of the assumptions proves incorrect, the results of salesforce.com, inc. could differ materially from the results expressed or implied by the forward-

looking statements we make. All statements other than statements of historical fact could be deemed forward-looking, including any projections of

product or service availability, subscriber growth, earnings, revenues, or other financial items and any statements regarding strategies or plans of

management for future operations, statements of belief, any statements concerning new, planned, or upgraded services or technology developments

and customer contracts or use of our services.

The risks and uncertainties referred to above include – but are not limited to – risks associated with developing and delivering new functionality for our

service, new products and services, our new business model, our past operating losses, possible fluctuations in our operating results and rate of growth,

interruptions or delays in our Web hosting, breach of our security measures, the outcome of intellectual property and other l itigation, risks associated

with possible mergers and acquisitions, the immature market in which we operate, our relatively limited operating history, our ability to expand, retain,

and motivate our employees and manage our growth, new releases of our service and successful customer deployment, our limited history reselling

non-salesforce.com products, and utilization and selling to larger enterprise customers. Further information on potential factors that could affect the

financial results of salesforce.com, inc. is included in our annual report on Form 10-Q for the most recent fiscal quarter ended July 31, 2012. This

documents and others containing important disclosures are available on the SEC Filings section of the Investor Information section of our Web site.

Any unreleased services or features referenced in this or other presentations, press releases or public statements are not currently available and may

not be delivered on time or at all. Customers who purchase our services should make the purchase decisions based upon features that are currently

available. Salesforce.com, inc. assumes no obligation and does not intend to update these forward-looking statements.

Vijay Devadhar

Developer

Message Queues – what are they?

Asynchronous job queue infrastructure in salesforce.com server

Engine behind

• Dashboards

• Reports,

• Batch/Async Apex

• Bulk API

• and many more…

Message Queues – what are our volumes?

Averages about 60 million messages a day

Biggest instances account for 10 million messages a day

95 percentile for dequeue latency is 10 minutes

Message Queues – session description

Discuss scaling techniques used to

• Manage capacity

• Resource allocation

and lessons learned..

Visit with Princess Aurora

Average wait time – 90 mins

Average time with princess – 120 seconds

What you remember the most after a year? The long……… wait.

Visit with Goofy

Average wait time – none

Average time with goofy – 120 seconds or until you get bored

Meanwhile…..

Characters are not Created Equal

If Only We Hired More Princess Aurora

What can we do better?

Fire Goofy(s)

Hire more Princesses

Convert Goofy to Princess and vice-versa when needed

Swap Goofy for Princess Aurora

Message Queues Amusement Park

300 + rides and characters

Traffic which ebbs and flows with time of day, day of the week etc.,

Plenty, but finite set of resources available

Goal

Reduce wait times

Fairly allocate resources

Adapt to varying traffic patterns

Solution

A large Shared thread pool – No ride specific silos

Round robin the process of picking work

If world wants Dashboards, do Dashboards

Message Queue Real Time Latencies

Unlike Disneyland, each job takes variable amount of time

Wait time prediction is not accurate at the tail of the queue

We report real time and act on them if needed

Elastic Thread Pools

Can grow from initial size

This allows growth as traffic demands

Wait times feedback to thread pool grow, shrink decisions

Let’s Do a Puzzle

A man has

• One Sheep

• One Tiger

• One bundle of Grass

• One small boat

and a big River to cross…

Similar Puzzle with Messages

Forecasting has

• Several sales reps whose forecasts need update

• Forecast update for sales rep should also update VP of sales

• Multiple sales reps for same VP

and a big sales projection to put out ….

Solution

Browse and Cache

Pick up work on which you can obtain mutex lock

Jump ahead if needed

And in Real World..

Cache capacity is tuned to typical traffic pattern

At times cache can fill up

Messages may be escorted to the back of the queue

Bread Lines Vs. Turkey Lines

Same set of ovens baking both

Bread is the basic need, Turkey when ovens are free

If bread lines build up, stop cooking turkey

If no one wants bread, just give all ovens to cooking turkey

User-facing vs Background jobs

Same set of servers for both

Users need fast response; Background can wait

If user requests pile up, stop processing background

If no user requests, just process background jobs

But how do you make Turkeys stop?

Traffic lights

Measure key resources

When resource usage crosses threshold, slow down on background

Sensitive to CPU, Memory, I/O, connection usage

Lessons in Real World

Traffic types vary, traffic volumes vary

Handlers misbehave, components have bugs

Distributed systems scale very well, not if you need mutex.

Real time alerting, trending, traffic isolation, troubleshooting are

necessary