Scale your Node App, Forget the Infrastructure - Next Steps with Lambda

Serverless, and specifically AWS Lambda, is the new hotness. Getting started is pretty easy, but getting past the basics can be a bit rough. This session goes past the getting started stage and continues with real life architectural learnings about how to make amazing applications on Lambda.

Transcript

1. Scale your Node Application, Forget the Infrastructure Taking the Next Step with AWS Lambda Matthew Williams Evangelist @ Datadog mattw@datadoghq.com | @technovangelist DevOpsDays Boston Organizer
2. The plan § Introduction § Coding § Debugging § Orchestration § Lambda Setup § Deployment § Monitoring
3. Introduction to Lambda • Key Concepts • Serverless • Stateless • Trigger-based • What a Lambda Function is • Your executable code (Nodejs 4.3/6.10, C#, Java, Python) • No restrictions • Priced based on execution time and memory
4. Available Lambda Triggers • S3 • DynamoDB • Kinesis Streams • Simple Notification Service • Simple Email Service • Cognito • CloudFormation • CloudWatch Logs • CloudWatch Events • CodeCommit • Scheduled Events (powered by CloudWatch Events) • Echo • Lex • API Gateway • Invoked via the AWS CLI/APIs
5. Other Lambda Triggers • Zapier • Twilio • Others?
19. Coding
20. SOA/Microservices One function - one purpose Fine grained
21. Scalable Inputs and Outputs Lambda is very scalable Is your database as scalable
22. Check if data exists Most probably stateless BUT…container could be reused Check if data exists in /tmp (500MB) If using a database, look into context.callbackWaitsForEmptyEventLoop
23. Think about error handling DLQ SNS or SQS
24. Debugging
25. Use Modules with Local Harness • All the logic should be in Modules • The Handler should just be a minimal wrapper • Why? • Easier to run tests on the code • Easier to run locally with a local harness
29. AWS X-Ray
31. Orchestration
32. Universal Event Stream
33. Step Functions
34. Lambda Setup
35. Minimal but Specific Security • Your IAM roles should allow the minimum privileges • If you only need to read, don’t also allow write
36. Choosing the right memory • As you increase memory, you get more CPU
38. Choosing the right timeout • Stick with the default unless you need more
39. Deployment
40. Frameworks node-lambda
41. CloudFormation + SAM
42. Tagging
43. Monitoring
44. What are the Lambda Metrics aws.lambda.duration, min, max, sum aws.lambda.errors aws.lambda.invocations aws.lambda.iterator_age aws.lambda.throttles aws.lambda.deadlettererrors
45. Three options for monitoring Lambda 1. Add a custom metric to CloudWatch 2. Add metric directly to monitoring application 3. Add a line to the CloudWatch logs
49. Scale your Node Application, Forget the Infrastructure Taking the Next Step with AWS Lambda Matthew Williams Evangelist @ Datadog mattw@datadoghq.com | @technovangelist DevOpsDays Boston Organizer