AWS Lambda Deployments: Best Practices and Common Mistakes O'Reilly Software Architect's Conference 2016-11-14

AWS Lambda Deployments
Best Practices and Common Mistakes
Given by Derek C. Ashmore
November 14, 2016
©2016 Derek C. Ashmore, All Rights Reserved 1

Who am I?
• Professional Geek
since 1987
• Java/J2EE/Java EE
since 1999
• Cloud since 2012
• Roles include:
• Architect
• Developer
• Project Manager
• DBA
• System Admin

Discussion Resources
• This slide deck
– http://www.slideshare.net/derekashmore
• Sample code on my Github
– https://github.com/Derek-Ashmore/
• Sample Java AWS Lambda Source
– https://github.com/Derek-Ashmore/AWSLambdaExamples
• Slide deck has hyper-links!
– Don’t bother writing down URLs

Agenda
The
“What”
and “Why”
of AWS
Lambda
Developing
Lambda
Deploying
Lambda
Tales from
the Field
Summary /
Q&A

What are AWS Lambdas?
• You provide custom code -> AWS runs it
– Java, Node.js, Python
• Computing power with less management
– AWS manages the hardware
– AWS autoscales that hardware
– AWS maintains that hardware
• Lambdas are event driven
– API Gateway (e.g. RESTful Web Service call)
– Many more
• Lambdas are stateless
• Not to be confused with “Lambda Expressions” in Java 8

Lambda Event Sources
• API Gateway
• SNS Messaging
Subscriptions
• Schedule
• Storage writes
– S3, DynamoDB, Kenesis
• SES Email receipt
• Cloudwatch
– Schedule, Events, log entries
• Cognito (Security)
• CloudFormation
– Creation script

What’s the Business Benefit
• Less Maintenance Hassle
• Unlimited Parallelism
• Current cost advantage
– Don’t pay for idle
– CPU cost currently lower
• Free tier
– 1 M executions and 400K compute seconds per month
– Memory allocated determines allowed free-tier runtime
• 20 cents per 1 M executions + memory/runtime cost
– Administration cost
• No O/S upgrades, server backups, etc.

There’s no free lunch
• Less control over environment
– Harder to tune
– Memory and time limits on execution
• Few Environment amenities
– No connection pooling, session support, caching
• Proprietary Interface
– Potential Technical Lock-in
• No Guarantee that AWS cost will be constant
– Potential Business Risk
• Modern version of CGI

Lambda Competitors
• Azure Functions (here)
– Closest matching feature set
• Large number of event types
• Node.js and C# Language Support
– Claim to support more, but not documented
• Pricing Model Similar
• Better developer support (IDE integrated with portal)
• Google Cloud Functions (here)
– Still Alpha
– Fewer event types
– Node.js language support

Agenda
The
“What”
and “Why”
of AWS
Lambda
Developing
Lambda
Deploying
Lambda
Tales from
the Field
Summary /
Q&A

Lambda API
• Create a Request Handler
– Inputs are
• Event input (user input arguments)
• Context input (info about execution environment)
– Outputs are user decided
• JSON marshalling of Request and Response
• Execution Context
• Deployment Zip Artifact
– Typically one RequestHandler per Zip artifact

Lambda Request / Response Sample
• Expose REST API that collects email addresses
for a mailing list
– https://scrubbed/prod/EmailCollector
• Sample request/response

Java Request Handler Example
• RequestHandler interface is generic.
– POJOs represent the request and response
– POJOs determine JSON request/response format
– Execution Context class provided by AWS
• This class specified when Lambda defined
• Note the wrapping try/catch
– Portion of the Context provided by AWS -> need more
– AWS Does marshalling
– ContextedRuntimeException from Apache Commons Lang3
• Note that the Lambda is thin -> Business logic is elsewhere
• Sample is on GitHub (here)
• Node.js and Python similar

What’s in the Context?
• Execution Context provided by AWS
• Contains:
– AWS Request ID -> Get logs for specific request
– Function name, version, arn
– Cognito identity
– Remaining time/memory
– Mobile client information (AWS Mobile SDK)
• Environment name/value map
• Custom name/value map

Resource Support
• Runtime Support
– you can configure memory and max run time available
– Java  Java 8; AWS SDK jar must be included in your deployment
– Node.js  AWS SDK for Node.js preinstalled
– Python  AWS SDK for Python preinstalled
• All logging viewable/searchable in Cloudwatch logs
– Node.js and Python log to the console
– Java uses customer Log4J Appender
• Third party jars/libraries can be included
– Including AWS SDK for AWS Tasks (executing other Lambdas)
– Database Drivers
– Web service libraries
• Keep in Mind
– You create/destroy all database connections
• No connection pooling (Lambdas are stateless)
– Caching APIs have limited benefit (Lambdas are stateless)
– No Remote Debug capability

Agenda
The
“What”
and “Why”
of AWS
Lambda
Developing
Lambda
Deploying
Lambda
Tales from
the Field
Summary /
Q&A

Lambda Deployment Package
• Zip Organization (Java example)
– Root is in the classpath
– Lib contains 3rd party jars
• Maven example here
– Need maven-dependency-plugin, maven-antrun-plugin

API Gateway
• Exposes Lambdas as a RESTful Web Service
– Can be publicly available or part of a secured
private microservice library

Gateway Lambda Integration
• Integrations do basic transformations
– Map headers and parameters to Lambda request
fields

Gateway Integration Example

Lambdas and SNS Topics
• Lambdas can subscribe to SNS Publish/subscribe
topics
• Request Message is type SNSEvent

Lambdas can be scheduled
• Lambda executions can be scheduled through
CloudWatch
– Cron expressions supported

Agenda
The
“What”
and “Why”
of AWS
Lambda
Developing
Lambda
Deploying
Lambda
Tales from
the Field
Summary /
Q&A

Chief Complaints
• Documentation leaves a lot to be desired
– This is an understatement.
– For Java Lambdas, you are almost on your own.
• Lambda start-up time not consistent
– Sometimes long start-up time for JVM
– Python is the fastest
• Optimizations that depend on state aren’t as easy
– You would have to persist that state
• This would have it’s own concurrency and performance issues

Implementation Tips
• Separation of Concerns
– Keep Lambda code separate from business logic
• Might want to change vendors someday
– Keep AWS SDK code separate from business logic
• Same reason
– Invoke other Lambdas through the API Gateway, not directly through the AWS SDK
• Same reason
– Keep Business Logic locally runnable/debuggable
• Remote debug isn’t yet possible
• Ensure you can always tie AWS Request Id to your business transaction
– Need a way to gather logs from a complete business transactions and the many services it might use
– All invocations get unique AWS request Ids
• For example, lambda invokes other lambdas
– Configure log4j layout (Java) to include AWS Request Id (example)
– Node.js and Python logs have request id automatically

Common Use Cases
• Processing for uploaded data
– Image processing
• Low-volume web site features
– Paying for idle cost prohibitive
• Scheduled Batch Work
– Break up batch by invoking other Lambdas
– The resulting scaling is delegated to AWS

Performance
• Start-up Time – Python, Node.js, Java
– Berezovsky performance test
• Throughput – Java and Node.js
– Both have a JIT
– DZone comparison here
– Close enough that other factors would likely guide your language
choice

Lambdas and Microservices

Using Lambdas as Microservices
• Lambda / API Gateway is a deployment option for microservices
– No differences in design principles
• Single purpose
• Self-contained
– Still design for failure
• Don’t “assume” that Lambda outages can’t happen
– A Lambda might need external resources that aren’t available
• Off Limits: Coding patterns that use state
– Lambdas must be stateless
– Fail fast patterns
• Service Call Mediator
• Circuit Breaker
– Performance Patterns
• Expiring Cache (API Gateway allows request caching)

Lambda and the Gartner Hype Cycle

Further Reading
• This slide deck
– http://www.slideshare.net/derekashmore
• AWS Lambda Reading List
– http://www.derekashmore.com/2016/04/aws-lambda-reading-list.html

Questions?
• Derek Ashmore:
– Blog: www.derekashmore.com
– LinkedIn: www.linkedin.com/in/derekashmore
– Twitter: https://twitter.com/Derek_Ashmore
– GitHub: https://github.com/Derek-Ashmore
– Book: http://dvtpress.com/

AWS Lambda Deployments: Best Practices and Common Mistakes O'Reilly Software Architect's Conference 2016-11-14

More Related Content

What's hot

Similar to AWS Lambda Deployments: Best Practices and Common Mistakes O'Reilly Software Architect's Conference 2016-11-14

Recently uploaded

AWS Lambda Deployments: Best Practices and Common Mistakes O'Reilly Software Architect's Conference 2016-11-14