Building Scalable Microservices with Kubernetes, gRPC, and Containers - API World 2015

Video: https://www.youtube.com/watch?v=UOIJNygDNlE

Transcript

1. @sandeepdinesh Scalable Microservices gRPC and Kubernetes and Containers oh my! Sandeep Dinesh, Cloud Developer Advocate
2. @sandeepdinesh Have you heard of our lord and savior, Microservices?
3. @sandeepdinesh Google Trends: Microservices
4. @sandeepdinesh “Write programs that do one thing, and do it well. Write programs that work together” - Unix Philosophy UNIX
5. @sandeepdinesh Why Microservices? Scalability Security Polyglot Development Maintainability Continuous Deployment Decentralization Robustness
6. @sandeepdinesh Why Microservices? Scalability Security Polyglot Development Maintainability Continuous Deployment Decentralization Robustness How?
8. Everything at Google runs in containers We launch over 2 billion containers per week. • Gmail, Web Search, Maps, ... • MapReduce, batch, ... • GFS, Colossus, ...
9. @sandeepdinesh Kubernetes Borg Stubby
10. @sandeepdinesh Kubernetes
11. @sandeepdinesh App A
12. @sandeepdinesh App A
13. @sandeepdinesh Replication Controller App A
14. @sandeepdinesh Replication Controller App A App A App A App A
15. @sandeepdinesh Replication Controller Service A App A App A App A App A
16. @sandeepdinesh Microservice A
17. @sandeepdinesh That’s it!
18. @sandeepdinesh Microservice A Microservice C Microservice D Microservice B
19. @sandeepdinesh Microservice A Microservice C Microservice D Microservice B
20. @sandeepdinesh Microservice A Microservice C Microservice D Microservice B
21. @sandeepdinesh Microservice A Microservice C Microservice D (2.0) Microservice B
22. @sandeepdinesh Let’s build
23. @sandeepdinesh JavaScript var http = require('http'); var server = http.createServer(function (request, response) { response.writeHead(200, {"Content-Type": "text/plain"}); response.end("Hello World\n"); }); server.listen(3000); App
24. @sandeepdinesh Bash $ docker build -t api-world/frontend:1.0 . $ docker tag api-world/frontend:1.0 gcr.io/<project>/frontend:1.0 $ gcloud docker push gcr.io/<project>/frontend:1.0 Build the Container
25. @sandeepdinesh YAML apiVersion: v1 kind: ReplicationController metadata: labels: name: frontend name: frontend spec: replicas: 2 selector: name: frontend Frontend Replication Controller template: metadata: labels: name: frontend spec: containers: - image: gcr.io/project-id/frontend:1.0 name: frontend ports: - containerPort: 3000 name: http-server
26. @sandeepdinesh Bash $ kubectl create -f frontend-controller.yaml Launch the Controller
27. @sandeepdinesh YAML apiVersion: v1 kind: Service metadata: name: frontend labels: name: frontend spec: type: LoadBalancer Frontend Service ports: - port: 80 targetPort: 3000 protocol: TCP selector: name: frontend
28. @sandeepdinesh Bash $ kubectl create -f frontend-service.yaml Launch the Service
29. @sandeepdinesh
30. @sandeepdinesh What about Communication?
31. @sandeepdinesh Microservice A Microservice C Microservice D Microservice B
32. @sandeepdinesh “The biggest issue in changing a monolith into microservices lies in changing the communication pattern.” - Martin Fowler
33. O(1010) RPCs per second at Google
34. @sandeepdinesh
35. @sandeepdinesh Protocol Buffers
36. message HelloRequest { string greeting = 1; } message HelloResponse { string reply = 1; } service HelloService { rpc SayHello(HelloRequest) returns (HelloResponse); } IDL (Interface definition language) Describe once and generate interfaces for any language. Data Model Structure of the request and response. Binary format for network transmission. Wire Format Protocol Buffers
37. @sandeepdinesh Protocol Buffers HTTP/2
38. @sandeepdinesh HTTP/2 HTTP/1.1 http://www.http2demo.io/
39. Multiplexing Bidirectional Streaming Flow Control HTTP/2 No more polling, sockets, or clunky SSE Control your congestion Single TCP connection
40. @sandeepdinesh Protocol Buffers HTTP/2 Multiple Languages
41. Multiple Languages C/C++
42. @sandeepdinesh Protocol Buffers HTTP/2 Multiple Languages Mobile First
43. @sandeepdinesh Let’s upgrade our app
44. @sandeepdinesh ProtoBuff package geo; service GeoService { rpc DistanceBetween (Points) returns (Distance) {} } message Point { float lat = 1; float lng = 2; } message Points { Point origin = 1; Point destination = 2; } message Distance { float distance = 1; } .proto
45. @sandeepdinesh JavaScript var grpc = require('grpc'), proto = grpc.load('interface.proto'), GeoService = grpc.buildServer([proto.geo.GeoService.service]); Backend Code
46. @sandeepdinesh JavaScript var server = new GeoService({ 'geo.GeoService': { distanceBetween: function(call, callback) { callback(null, getDistance(call.request)); } } }); Backend Code
47. @sandeepdinesh JavaScript function distance(a,b,c,d){//haversine formula stuff} function getDistance(points){ return distance(points.origin.lat, points.origin.lng, points.destination.lat, points.destination.lng); } Backend Code
48. @sandeepdinesh JavaScript server.bind('0.0.0.0:50051'); server.listen(); Backend Code
49. @sandeepdinesh Bash $ docker build -t api-world/backend:1.0 . $ docker tag api-world/backend:1.0 gcr.io/<project>/backend:1.0 $ gcloud docker push gcr.io/<project>/backend:1.0 Build the Container
50. @sandeepdinesh JavaScript var grpc = require('grpc'), proto = grpc.load('interface.proto'), client = new proto.geo.GeoService('backend:50051'); Frontend Code
51. @sandeepdinesh JavaScript var request = { origin: { lat: 0, lng: 0 }, destination: { lat: 15, lng: 15 } } Frontend Code
52. @sandeepdinesh JavaScript client.distanceBetween(request, function(error, distance) { if (error) { response.end(JSON.stringify(error)); } else { response.end("Distance = " + JSON.stringify(distance) + "\n"); } }); Frontend Code
53. @sandeepdinesh Bash $ docker build -t api-world/frontend:2.0 . $ docker tag api-world/frontend:2.0 gcr.io/<project>/frontend:2.0 $ gcloud docker push gcr.io/<project>/frontend:2.0 Build the Container
54. @sandeepdinesh YAML apiVersion: v1 kind: ReplicationController metadata: labels: name: backend name: backend spec: replicas: 4 selector: name: backend Backend Replication Controller template: metadata: labels: name: backend spec: containers: - image: gcr.io/project-id/backend:1.0 name: backend ports: - containerPort: 50051 name: grpc-server
55. @sandeepdinesh Bash $ kubectl create -f backend-controller.yaml Launch the Controller
56. @sandeepdinesh YAML apiVersion: v1 kind: Service metadata: name: backend labels: name: backend spec: Backend Service ports: - port: 50051 targetPort: 50051 selector: name: backend
57. @sandeepdinesh Bash $ kubectl create -f backend-service.yaml Launch the Service
58. @sandeepdinesh Bash $ kubectl rolling-update frontend \ --image=gcr.io/smart-spark-93622/frontend:2.0 \ --update-period="5m" Upgrade the Frontend
59. @sandeepdinesh
60. @sandeepdinesh Kubernetes k8s.io grpc.io
61. @sandeepdinesh Google Container Engine
62. Thank you! Thank you! k8s.io grpc.io cloud.google.com https://github.com/thesandlord/samples/tree/master/grpc-kubernetes-microservices @sandeepdinesh
63. @sandeepdinesh Kubernetes Cluster Applications Ops Dev App A App B App C App A