Continuous verification for serverless applications - AWS Community Day Amsterdam October 27 2020

Presented at AWS Community Day Amsterdam, October 27, 2020.

@gunnargrosch
failure-lambda

The transition into more complex systems is accelerating and the granularity in serverless applications makes even smaller systems complex and highly distributed. Chaos engineering draws from the rich history of empirical experimentation to proactively discover vulnerabilities in these complex systems and help us verify whether or not the system behaves as expected under adverse conditions. Join as we raise the bar for how chaos engineering can be used with serverless applications and help us to continuously verify the output of the system through automation and the advantages established by CI/CD.

Transcript

1. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Gunnar Grosch @gunnargrosch October 27, 2020 Continuous verification for serverless applications AWS Community Day Amsterdam
2. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. “Testing to ensure that you can meet your availability goals is the only way you can have confidence that you will meet those goals” Reliability Pillar AWS Well-Architected Framework
3. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Verification of applications
4. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Verification of applications Objectives Verify Measures Conditions
5. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Verification of applications Objectives Verify Measures Conditions
6. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Conditions for verification Errors in your code Security policy errors Service configuration errors Function disk space failure Downstream service issues Concurrency and throttling Latency Client Amazon Simple Storage Service (Amazon S3) Amazon API Gateway AWS Lambda Amazon DynamoDB AWS Lambda Amazon Simple Storage Service (Amazon S3)
7. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Tools for adding conditions Chaos-lambda Python Failure-lambda NodeJS
8. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Failure-lambda NodeJS NPM package for NodeJS Lambdas Configuration using Parameter Store or AWS AppConfig Several failure modes • Latency • Status code • Exception • Disk space • Denylist const failureLambda = require('failure-lambda’) exports.handler = failureLambda(async (event, context) => { ... }) { "isEnabled": false, "failureMode": "latency", "rate": 1, "minLatency": 100, "maxLatency": 400, "exceptionMsg": "Exception message!", "statusCode": 404, "diskSpace": 100, “denylist": [ "s3.*.amazonaws.com", "dynamodb.*.amazonaws.com" ] }
9. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Verification of applications Objectives Verify Measures Conditions
10. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Verification example Objective: My purchase API should respond in less than 400 ms Measure: 400 ms threshold Condition: 100-200 ms latency injection to function Verify: Pass or fail Client Amazon API Gateway Amazon DynamoDB AWS Lambda
11. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Verification example Objective: I want 10 clicks per second Measure: 10 clicks threshold Condition: Inject downstream service failure Verify: Pass or fail Client Amazon API Gateway Amazon DynamoDB AWS Lambda
12. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Making verification continuous
13. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Making verification continuous
14. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Making verification continuous Objective: My purchase API should respond in less than 400 ms Measure: 400 ms threshold Condition: 100-200 ms latency injection to function Verify: Pass Default deploy
15. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Making verification continuous Objective: My purchase API should respond in less than 400 ms Measure: 400 ms threshold Condition: 100-200 ms latency injection to function Verify: Fail Canary deploy
16. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Making verification continuous Objective: My purchase API should respond in less than 400 ms Measure: 400 ms threshold Condition: 100-200 ms latency injection to function Verify: Pass Feature flag
17. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Key learnings Verification tells you why it’s important to explore a set of conditions It’s not about breaking things, it’s about learning and building confidence It’s easy to get started.
18. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. “Testing to ensure that you can meet your availability goals is the only way you can have confidence that you will meet those goals” Reliability Pillar AWS Well-Architected Framework
19. © 2020, Amazon Web Services, Inc. or its Affiliates. All rights reserved. Thank you! Gunnar Grosch @gunnargrosch