Running High Performance And Fault Tolerant Elasticsearch Clusters On Docker

06/07/2016 - 11:50 to 12:10
short talk (20 min)

Session abstract: 

We’ve learned how to run Docker containers. And we’ve learned how to run Elasticsearch. However, to run containerized Elasticsearch nodes and to do that effectively -- and at scale -- takes a little more knowledge and work. Sure, containers can be easily started and stopped; but how do you do that with Elasticsearch inside them?

In this talk we’ll quickly run over the basic Docker+Elasticsearch setup and focus on harder problems:

  • Architecting for Elasticsearch fault tolerance and high availability in containerized setup - using sharding, replication, node and shard-awareness for keeping your cluster green
  • Running Elasticsearch in different modes with re-usability in mind
  • Optimizing and tuning Elasticsearch for popular use cases like ELK
  • Ops/Devops - monitoring Elasticsearch & Docker together - which metrics to watch, what they mean, how to act on them and first of all, how to watch them


  • Running Elasticsearch on Docker
  • Performance implications of running Elasticsearch on Docker
  • How to create highly available and fault tolerant clusters
  • Proper Elasticsearch setup in a container
  • Tuning Elasticsearch
  • Which metrics should be looked at when dealing with Elasticsearch
  • Monitoring containers and applications running inside of them