Measure and Increase Developer Productivity with Help of Serverless at Serverless Architecture Conference Berlin 2020
- 1. Measure and increase developer productivity with the help of Serverless by Christian Bannes and Vadym Kazulkin, ip.labs GmbH, 14.10.2020
- 2. Contact Vadym Kazulkin ip.labs GmbH v.kazulkin@iplabs.de @VKazulkin @ServerlessBonn (Meetup) Christian Bannes ip.labs GmbH c.bannes@iplabs.de xing.com/profile/Christian_Bannes https://www.linkedin.com/in/vadymkazulkin/ https://www.iplabs.de/
- 4. Let’s talk about the challenges of the software development in general first
- 5. Cognitive Load – the total amount of mental effort being used in the working memory • Intrinsic • Extraneous • Germane https://teamtopologies.com/
- 6. Cognitive Load • Intrinsic • How to write a Java class or use a framework (Spring) • Extraneous • Germane
- 7. Cognitive Load • Intrinsic • Extraneous • How to automate tests (unit, integration, end-to-end, web, desktop, mobile) • How to build, package, deploy and run my application • How to configure monitoring, alerting, auto-scaling, logging and tracing • How to operate and maintain infrastructure • How to build-in fault-tolerance and resiliency • How to make the hardware, networking and application secure • Germane
- 8. Cognitive Load • Intrinsic • Extraneous • Germane • Domain Knowledge (payment, e-commerce) • Business processes and workflows
- 9. Cognitive Load • Intrinsic -> become fluent in it • Extraneous -> minimize amount of what we implement/operate/support/own by ourselves • Germane -> minimize amount of what we have to implement by ourselves
- 10. Productivity We are productive if we regularly ship products, which are successfully used by our customers
- 11. What is holding us back from being productive? Technical Debt - reflects the implied cost of additional rework caused by choosing an easy (limited) solution now instead of using a better approach that would take longer ”The Cost of Poor Quality Software in the US: A 2018 Report” https://www.it-cisq.org/the-cost-of-poor-quality-software-in-the-us-a-2018-report/The-Cost-of-Poor-Quality-Software-in-the-US-2018-Report.pdf
- 12. Technical Debt • Even a perfect solution can become the technical debt over the time • Version of programming language comes out of support (Java 8, JS Frameworks) • Security considerations forces us to upgrade one of our dependencies (library or web application server version) • One of our dependencies (e.g. to open source project) is discontinued
- 13. Technical Debt Think of what can happen to your software over the entire life cycle of our product
- 14. Technical Debt • is related to amount of code written • is related to amount of dependencies used • open source projects, programming languages, databases, (web) application servers
- 15. Legacy Systems are systems that can’t evolve ”The Cost of Poor Quality Software in the US: A 2018 Report” https://www.it-cisq.org/the-cost-of-poor-quality-software-in-the-us-a-2018-report/The-Cost-of-Poor-Quality-Software-in-the-US-2018-Report.pdf Legacy System
- 16. Evolutionary Architecture – supports guided, incremental change across multiple dimensions • Incremental change • Appropriate architectural coupling • Fitness functions ”Architectural Coupling” https://learning.oreilly.com/library/view/building-evolutionary-architectures/9781491986356/ch04.html
- 17. Evolutionary Architecture – Fitness functions • Source code metrics (such as measuring cyclomatic complexity) • Unit tests (% of coverage and % of success) • Performance metrics (such as API latency or throughput) • Security (encryption at rest, e.g. checking that all S3 buckets have encryption enabled, or automatic key rotation for all external APIs, with tools such as the AWS Secrets Manager) • ArchUnit, Sonar, CI/CD Tools Danilo Poccia „ Serverless + Evolutionary Architectures + Safe Deployments = Speed in the Right Direction” https://blog.usejournal.com/serverless-evolutionary-architectures-safe-deployments-speed-in-the-right-direction-7b4b01e27254
- 18. The Value Proposition of Serverless But let’s talk about of Total Cost of Ownership of the Serverless paradigm
- 19. TCO Full Picture No Infrastructure Operation and Maintenance Forrest Brazeal „The Business Case For Serverless” https://www.trek10.com/blog/business-case-for-serverless
- 20. No Infrastructure Maintenance Is infrastructure maintenance and operation your core competency ?
- 21. TCO Full Picture No Infrastructure Operation and Maintenance Auto Scaling and Fault Tolerance Built in Forrest Brazeal „The Business Case For Serverless” https://www.trek10.com/blog/business-case-for-serverless
- 22. Auto Scaling And Faul Tolerance Built In • Can you get capacity planning and auto scaling right? • Do you want to solve the hard problem of fault tolerance by yourself?
- 23. TCO Full Picture No Infrastructure Operation and Maintenance Auto Scaling and Fault Tolerance Built in Do more with less Forrest Brazeal „The Business Case For Serverless” https://www.trek10.com/blog/business-case-for-serverless
- 24. Do more with less By heavily relying on the managed Serverless services you • Need fewer engineers to start implementing your new product idea • Can do more with the same amount of people
- 25. TCO Full Picture No Infrastructure Operation and Maintenance Auto Scaling and Fault Tolerance Built in Do more with less Lower technical debt Forrest Brazeal „The Business Case For Serverless” https://www.trek10.com/blog/business-case-for-serverless
- 26. Lower technical debt • Whatever code you write today is always tomorrow’s technical debt © Paul Johnston • Less code means lower technical debt • Time and effort required for maintaining the solution over its whole lifecycle is by far much more than for developing it Jeff Atwood „The Best Code is No Code At All” https://blog.codinghorror.com/the-best-code-is-no-code-at-all/ Paul Johnston “Cloud 2.0: Code is no longer King — Serverless has dethroned it” https://medium.com/@PaulDJohnston/cloud-2-0-code-is-no-longer-king-serverless-has-dethroned-it-c6dc955db9d5
- 27. TCO Full Picture No Infrastructure Operation and Maintenance Auto Scaling and Fault Tolerance Built in Do more with less Lower technical debt Focus on Business Value and Innovation Forrest Brazeal „The Business Case For Serverless” https://www.trek10.com/blog/business-case-for-serverless
- 28. Focus On Business Value and Innovation Every organization wants exactly this!
- 29. TCO Full Picture No Infrastructure Operation and Maintenance Auto Scaling and Fault Tolerance Built in Do more with less Lower technical debt Faster Time to Market Forrest Brazeal „The Business Case For Serverless” https://www.trek10.com/blog/business-case-for-serverless Focus on Business Value and Innovation
- 30. Faster Time To Market • Time To Market is the key differentiator in today’s business! • Ask yourself: what is core for your business and what you can get as Commodity +(Utility) as a Service?
- 31. Serverless Mindset at ip.labs “Accelerate Innovation and Maximize Business Value with Serverless Applications” https://www.slideshare.net/AmazonWebServices/accelerate-innovation-and-maximize-business-value-with-serverless-applications-srv212r1-aws-reinvent-2018
- 32. Serverless Mindset at ip.labs If AWS offers a managed service with acceptable price and „good enough“ feature set, use it • AWS-native aka Serverless is your first option. • For the price consider the Total Cost of Ownership (TCO) If market offers a managed service with acceptable price and „good enough“ feature set, buy/rent it • PagerDuty instead of CloudWatch Alarm • Akamai or Digital Ocean CDN instead of AWS CloudFront If you have to manage it by yourself, manage it • Prometheus/Grafana instead of AWS CloudWatch If you can reconsider requirements, do it • Narrow the scope, get 80% of what you need from AWS or market, instead of building or hosting it yourself • Get faster time to market, don‘t lose the focus • AWS AppConfig instead of own Feature Toggle solution • AWS Step Functions instead of Camunda BPMN If you have to build it, own it • Business logic in the core domains Jared Short https://twitter.com/shortjared/status/1100887501047132160?lang=de
- 33. How to measure success See DORA State of DevOps 2018-2019 Reports
- 34. Software Delivery and Operational Performance See DORA State of DevOps 2018-2019 Reports
- 35. Example: strategies to reduce time to restore service See DORA State of DevOps 2018-2019 Reports
- 36. Blue-Green deployment “Win-Win Deployment Strategies for Modern Apps” https://rollbar.com/blog/deployment-strategies/ Canary deployment
- 37. AWS Lambda Deployment Best Practices: • API Gateway Stage variables and Lambda Aliases • Lambda Alias Traffic Shifting Chris Munns: “Testing and Deployment Best Practices for AWS Lambda-Based Applications” https://www.youtube.com/watch?v=zJQDAsWm-5k&list=LLYgjRSI2oCzI9eooyFrWR7A&index=1
- 38. AWS Lambda Deployment Best Practices • AWS Lambda Alias Canary and Linear Traffic Shifting & AWS SAM Safe Deployments • CloudWatch Rollback Alarms & Lambda hooks Chris Munns: “Testing and Deployment Best Practices for AWS Lambda-Based Applications” https://www.youtube.com/watch?v=zJQDAsWm-5k&list=LLYgjRSI2oCzI9eooyFrWR7A&index=1
- 39. Time Spent See DORA State of DevOps 2018-2019 Reports
- 40. Wardley Maps Simon Wardley https://www.slideshare.net/swardley/why-the-fuss-about-serverless-88107645 Co-evolution of practices to become productive with Serverless
- 41. Co-evolution of practices with Serverless 1/2 • True DevOps (even DevSecOps) • FinDev responsibilities in the teams • Complete infrastructure automation • Chaos Engineering Sheen Brisals “Why the ‘WHY’ matters more than the ‘WHAT’ in Serverless!” https://medium.com/lego-engineering/why-the-why-matters-more-than-the-what-in-serverless-2ef56c397962 DevOps Topologies: https://web.devopstopologies.com/
- 42. Co-evolution of practices with Serverless 2/2 • Each developer has its own (AWS test) account • No local testing environment or only for quick functional tests • Testing in production Michael Bryzek “What do you know about testing in production?” https://www.youtube.com/watch?v=z-ATZTUgaAo
- 43. Using Serverless ecosystem with the right engineering practices in place will significantly reduce • extraneous and germane cognitive load • the amount of code written
- 44. How to write less code with AWS Serverless services 1/3 • Write fewer Lambda functions • use direct AWS service integrations in case the Lambda only calls the AWS service itself
- 45. Less Lambda functions means less: • code to write, test, run and maintain • CI/CD (deploy, rollback strategies) to maintain • Infrastructure as a Code (IAM policies, permission) to write and test • cold-start worries • point of failures and retries • security concerns • worries about Lambda limits (e.g. concurrency settings per AWS account) • spending on Lambda, CloudWatch and 3rd party SaaS on (AWS) (monthly) bill Sheen Brisals “Don’t wait for Functionless. Write less Functions instead” https://medium.com/lego-engineering/dont-wait-for-functionless-write-less-functions-instead-8f2c331cd651
- 46. How to write less code with AWS Serverless services 2/3 • Write fewer Lambda functions • use more direct service integrations in case the Lambda only calls the Service itself • Write less Infrastructure as a Code • applies not only for Lambda but also to other services like AppSync • Use (AWS) Services/Frameworks which provide meaningful abstractions • Amplify Framework or Serverless Framework Components
- 47. How to write less code with AWS Serverless services 3/3 • API Gateway Service Integration • HTTP APIs Storage-First Service Integration • Step Functions Service Integration • Event Bridge Filtering and Routing • Lambda Destinations • AppSync and Direct Lambda Resolvers • Amplify Framework • Serverless Framework Components
- 48. Sheen Brisals “Don’t wait for Functionless. Write less Functions instead” https://medium.com/lego-engineering/dont-wait-for-functionless-write-less-functions-instead-8f2c331cd651 API Gateway Service Integration
- 49. API Gateway Service Integration with Dynamo DB : Example ID generator 1/2 Sheen Brisals “Sequence Numbering in Serverless via API Gateway” https://medium.com/lego-engineering/sequence-numbering-in-serverless-via-api-gateway-40e5f6c83e93 https://github.com/ToQoz/api-gateway-mapping-template
- 50. API Gateway Service Integration with Dynamo DB : Example ID generator 2/2 Sheen Brisals “Sequence Numbering in Serverless via API Gateway” https://medium.com/lego-engineering/sequence-numbering-in-serverless-via-api-gateway-40e5f6c83e93 https://github.com/ToQoz/api-gateway-mapping-template
- 51. Eric Johnson “https://aws.amazon.com/de/blogs/compute/building-storage-first-applications-with-http-apis-service-integrations/” HTTP APIs Storage-First Service Integration • Event Bridge • Kinesis Data Streams • SQS • AppConfig • Step Functions
- 52. Step Functions Service Integration “Service Integrations with AWS Step Functions” https://docs.aws.amazon.com/step-functions/latest/dg/concepts-service-integrations.html
- 53. Event Bridge Filtering and Routing “Reducing custom code by using advanced rules in Amazon EventBridge” https://aws.amazon.com/de/blogs/compute/reducing-custom-code-by-using-advanced-rules-in-amazon-eventbridge/
- 54. Lambda Destinations “Introducing AWS Lambda Destinations” https://aws.amazon.com/de/blogs/compute/introducing-aws-lambda-destinations/ https://www.trek10.com/blog/lambda-destinations-what-we-learned-the-hard-way
- 55. AppSync and Direct Lambda Resolvers https://aws.amazon.com/de/appsync “Introducing Direct Lambda Resolvers: AWS AppSync GraphQL APIs without VTL” https://aws.amazon.com/de/blogs/mobile/appsync-direct-lambda/
- 58. Using Serverless ecosystem will with the right engineering practices in place will significantly reduce • extraneous and germane cognitive load • the amount of code written • amount of dependencies (no execution runtime and web application server to take care of)
- 59. Serverless with the focus on you core domains will enable • iterative development mind set • experimentation culture • focus on business value innovation and faster time to market • evolutionary architectures
- 60. Evolutionary Architecture – supports guided, incremental change across multiple dimensions • Incremental change • Appropriate architectural coupling • Fitness functions ”Architectural Coupling” https://learning.oreilly.com/library/view/building-evolutionary-architectures/9781491986356/ch04.html • Serverless ecosystem supports evolutionary architectures really well • All Serverless solutions often suffer from the “Antipattern: Last 10% Trap”
- 61. AWS Serverless ecosystem Overview of the recent major improvements
- 62. Architecture and workloads types • Event-driven • API-driven • Batch Job • Internal Tool • ML/AI • Big Data Image: flickr.com/photos/everywhereatonce/294789504 Christian Bannes and Vadym Kazulkin @VKazulkin , ip.labs GmbH
- 63. Lambda Layers & Lambda Runtime API Door opener for use cases like: • Big Data • ML/AI Christian Bannes and Vadym Kazulkin @VKazulkin , ip.labs GmbH
- 64. A Shared File System for Your Lambda Functions Door opener for use case like: • ML/AI Christian Bannes and Vadym Kazulkin @VKazulkin , ip.labs GmbH
- 65. Provisioned Concurrency for Lambda Functions Christian Bannes and Vadym Kazulkin @VKazulkin , ip.labs GmbH
- 66. Lambda functions as targets for Application Load Balancers Christian Bannes and Vadym Kazulkin @VKazulkin , ip.labs GmbH
- 67. HTTP APIs in Beta 70% cheaper as API Gateway • Fewer configuration options • Well-suited for most use-cases
- 68. DynamoDB On-Demand Capacity mode Christian Bannes and Vadym Kazulkin @VKazulkin , ip.labs GmbH
- 69. Jeremy Daly: “Mixing VPC and Non-VPC Lambda Functions for Higher Performing Microservices” https://www.jeremydaly.com/mixing-vpc-and-non-vpc-lambda-functions-for-higher-performing-microservices/ Vadym Kazulkin @VKazulkin , ip.labs GmbH Lambda behind the Virtual Private Cloud (VPC)
- 70. Lambda behind the VPC • The network interface creation happens when Lambda function is created or its VPC settings are updated • Reduced additional cold start from approx. 10 seconds to below 1 second Chris Munns: "Announcing improved VPC networking for AWS Lambda functions” https://aws.amazon.com/de/blogs/compute/announcing-improved-vpc-networking-for-aws-lambda-functions/
- 71. Amazon Aurora Serverless Data API as beta for MySql and Postgres available https://docs.aws.amazon.com/de_de/AmazonRDS/latest/AuroraUserGuide/data-api.html Christian Bannes and Vadym Kazulkin @VKazulkin , ip.labs GmbH
- 72. RDS Proxy – Generally Available for Aurora MySQL, Aurora PostgreSQL, RDS MySQL and RDS PostgreSQL Christian Bannes and Vadym Kazulkin @VKazulkin , ip.labs GmbH
- 73. Compute Savings Plans now cover AWS Lambda Usage Christian Bannes and Vadym Kazulkin @VKazulkin , ip.labs GmbH
- 74. Recent CloudWatch Improvements • Search over multiple Log Groups became possible • CloudWatch Logs Insights • enables you to interactively search and analyze your log data in Amazon CloudWatch Logs • Embedded Metric Format • JSON specification used to instruct CloudWatch Logs to automatically extract metric values embedded in structured log events.
- 75. Further Improvement Areas for Serverless Ecosystem 1/2 • EFS improvements • Enable calling Lambda upon EFS Events (file created or updated) • Integrate with AWS compliance and governance services e.g. AWS Config, AWS CloudTrail like S3 does • CloudWatch improvements • Observability (no match to Lumigo or Epsagon) • Alarms (no match to PagerDuty or Lumigo or Epsagon)
- 76. Further Improvement Areas for Serverless Ecosystem 2/2 • CodeCommit improvements • not nearly comparable to GitHub and BitBucket • X-Ray support for all (async) Serverless services • EventBridge
- 77. Serverless in practice ( for large scale projects) Situation at ip.labs • Hybrid (still having monolithic java application) • One team completely serverless for two years • Another team just started using serverless only technology
- 79. Before we went serverless... • Central system administration team • Tickets prioritized • Lead to waiting time Waiting = WASTE
- 80. No low level administration • No need for low level administration at networking level • No servers and no OS • Less interaction with administration team
- 81. Responsibility Shift • Still need for operation • Serverless = Devops • Less waiting time means you go faster
- 83. Managed Services • Use managed services for generic subdomains • Example: Authentication with Cognito
- 84. Live Demo • Static webpage on S3 • API via Http API Gateway • Authentication via Cognito • JWT Authorizer
- 86. Architecture for serverless • 'just functions' - no need for architecture? • Architecture = long term investment • Important for large scale projects
- 87. Organize lambdas around business domain Domain
- 88. Organize lambdas around bussiness domain ... Subdomain ... Identity & Access
- 89. Organize lambdas around bussiness domain Product Portfolio Identity & Access ... ... Bounded Context Product Inventory Shop Integration
- 91. Product Inventory Bounded Context Bounded Context
- 92. Use Case = One Lambda Function Use Case
- 93. Communication via Events Communication via events
- 95. Namespacing Data • Each service operates only on its own data • HashKey with namespace prefix • Policy ensures service can only access it's own data • Fine grained access control
- 96. Subdomain = Independant Application Product Portfolio Identity & Access ... ... Product Inventory Shop Integration
- 98. Why you should separate • Evolve technology and business independently • Expect technology changes in the future • Refactor technology without changing business code
- 99. https://en.wikipedia.org/wiki/Hexagonal_architecture_(software) Ports & Adapters • Interface in language of your domain = ports • Implementation = adapters • Inject the implementation into your domain
- 100. Ports & Adapters implementation Adapters Ports
- 101. Ports & Adapters implementation Business method depens on interfaces Convert intput to domain object Adapters for injection
- 103. • Infrastructure as Code is a MUST • Create a CLI for scaffolding & common tasks Automation
- 104. • Creates serverless.yml with predefined settings • Project scaffolding • Deployment pipeline CLI usage cli create:service product-inventory
- 105. Continuous Delivery • Codepipeline for deployments • One Pipeline for one service • Every commit on master triggers deployment
- 107. Thank You!