Why Service is the worst API in Kubernetes, and what we can do about it

This was my KubeCon NA 2023 (Chicago) lightning talk.

Transcript

1. Google Cloud Platform Why Service is the worst API in Kubernetes, and what we can do about it KubeCon, Chicago Nov 6, 2023 Tim Hockin <thockin@google.com> @thockin
2. Google Cloud Platform “Service” is one of the oldest APIs in Kubernetes
3. Google Cloud Platform “Service” is one of the oldest APIs in Kubernetes $ git blame --ignore-rev bd7643c03339 pkg/apis/core/types.go | grep "type Service struct" ^2c4b3a562ce pkg/api/types.go (Joe Beda 2014-06-06 16:40:48 -0700 4358) type Service struct {
4. Google Cloud Platform “Service” is also one of the most widely used APIs in Kubernetes
5. Google Cloud Platform In-cluster virtual services
6. Google Cloud Platform In-cluster virtual services IP allocation
7. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers
8. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports
9. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS names & SRV
10. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS names & SRV Name aliases
11. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS names & SRV Name aliases LB health-checks
12. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS names & SRV Name aliases LB health-checks Routing policy
13. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS names & SRV Name aliases LB health-checks Routing policy Automatic endpoint management
14. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS names & SRV Name aliases LB health-checks Routing policy Automatic endpoint management Manual endpoints
15. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS names & SRV Name aliases LB health-checks Routing policy Automatic endpoint management Manual endpoints Session affinity
16. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS names & SRV Name aliases LB health-checks Routing policy Automatic endpoint management Manual endpoints Session affinity Node implementation hints
17. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS & SRV names Name aliases LB health-checks Routing policy Automatic endpoint management Manual endpoints Session affinity Node implementation hints Port mappings
18. Google Cloud Platform In-cluster virtual services IP allocation Out-of-cluster loadbalancers Node-ports DNS & SRV names Name aliases LB health-checks Routing policy Automatic endpoint management Manual endpoints Session affinity Node implementation hints Port mappings Simple firewall
19. Google Cloud Platform “Service” was designed to be simple! LoadBalancer NodePort ClusterIP
20. Google Cloud Platform As Kubernetes expanded and matured, we accumulated functionality: • Internal and external traffic policies • Dual-stack support • Topology awareness • LB Options • Don’t allocate NodePorts for LBs • Don’t allocate HCNPs
21. Google Cloud Platform The API that we laid out almost 10 years ago is starting to limit how we can evolve But, we have a strong commitment to compat, which includes under-specified semantics!
22. Google Cloud Platform Is session affinity per-service or per-port?
23. Google Cloud Platform Are implementations required to consider port-protocol or just port number when routing?
24. Google Cloud Platform Is a Service immutable, or can it be updated?
25. Google Cloud Platform Service API does too many things for too many use-cases Service API is different from other APIs in too many subtle ways • Example: synchronous IP and node-port allocation
26. Google Cloud Platform Result: A complex API to use and maintain • Lots of inter-related fields • Hard to validate and test • Hard to document
27. Google Cloud Platform Result: Hard to extend • “All ports” is basically impossible • Port naming across protocols is clunky • Implementations need more and more knobs • Adding different types of LBs is challenging
28. Google Cloud Platform So...what are we going to do about it?
29. Google Cloud Platform Gateway API
30. Google Cloud Platform Application Operator Infrastructure Provider Cluster Operator
31. Google Cloud Platform Gateway (front door) Application Operator Infrastructure Provider Cluster Operator
32. Google Cloud Platform Gateway (front door) Gateway Class (which impl) Application Operator Infrastructure Provider Cluster Operator
33. Google Cloud Platform Gateway (front door) Gateway Class (which impl) Application Operator Infrastructure Provider Cluster Operator *Route *Route *Route
34. Google Cloud Platform Gateway (front door) Gateway Class (which impl) Application Operator Infrastructure Provider Cluster Operator *Route *Route *Route *Route *Route Service
35. Google Cloud Platform Gateway class=ClusterIP
36. Google Cloud Platform Gateway class=LoadBalancer
37. Google Cloud Platform Gateway class=LoadBalancer Gateway class=ClusterIP
38. Google Cloud Platform Legacy model, evolved GW LoadBalancer class=internal Pod Selector Cluster IP Service Name GW ClusterIP Cluster IP Service Name GW LoadBalancer class=external
39. Google Cloud Platform This is not a commitment! • Several of these pieces are already in progress • Some are barely sketched out • Gateway API is hitting 1.0 imminently ◦ That doesn’t include ClusterIP support, yet I am seeking feedback on the idea!