Skip to content

A practical example of Redis Replication and High Availability mode with node js client in Kubernetes using correct way of handling statefull component in k8s i.e Statefulset.

sourikghosh/redisHA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

35 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

REDIS HIGH AVAILABILITY AND REPLICATION

This project mainly focuses on achieving HA and Replication on Kubernetes platform. I tried to follow all the necessary best practices unless and until it makes our workflow more complicated. Since our primary goal is to achieve HA.

CLIENT

I made a most simple and bare-bone client on nodejs to ensure our connection to Redis and for testing purposes. Since Redis is a key-value pair in-memory database (now it supports Json, Hash, etc) and just to make things simple I am setting a key and sending the response back to the client after fetching the key from the database.

For More info on client side and Kubernetes implementation visit Nodejs-Client

REDIS

Master-Replica replication

Use replicaof to make a Redis instance a copy of another Redis server. A few things to understand ASAP about Redis replication

  • Redis replication is asynchronous, but you can configure a master to stop accepting writes if it appears to be not connected with at least a given number of replicas

  • Redis replicas are able to perform a partial resynchronization with the master if the replication link is lost for a relatively small amount of time.

Challenges To achieve HA

  • Cannot really use Deployment to create Redis pods.Deployment are not meant for stateful components. There are many reasons for that.
  • Cannot use clusterIp service since its loadbalances the pod (in round-robin by default). Loadbalancing is not desired for Redis cluster because we only want to write on master and not on Replicas because writing on replicas will be discarded on the next sync with the master node
  • Cannot hard-coded the config file since replicaof field will keep changing in long-run.
  • How to deciede which will be the next master if the master dies.

Expected Solutions

  • We have to use Statefulset. Unlike a Deployment, a StatefulSet maintains a sticky identity for each of their Pods. These pods are created from the same spec, but are not interchangeable: each has a persistent identifier that it maintains across any rescheduling.
  • We have to use headless service. Kube-proxy does not handle these Services, and there is no load balancing or proxying done by the platform for them.For headless Services that define selectors, the endpoints controller creates Endpoints records in the API, and modifies the DNS configuration to return records (addresses) that point directly to the Pods backing the Servic.
  • One of the Solution is using initContainers to make the connection between replicas and master on the fly. Initcontainer spawns before the pod creation and makes the master-replica connection.
  • Redis solution to this is Redis Sentinels. Redis Sentinel provides high availability for Redis. In practical terms this means that using Sentinel you can create a Redis deployment that resists without human intervention certain kinds of failures.

To check all the Kubernetes command used kubernetes-command

Redis Persistence

Redis is a database though it's in-memory but we are enabling persistence in it by both append-only mode and rdb mode as well both the mode have its drawback.

About

A practical example of Redis Replication and High Availability mode with node js client in Kubernetes using correct way of handling statefull component in k8s i.e Statefulset.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published