Elasticsearch tune-up - Noteworthy at Compose

Published

This is your weekly summary of Compose news for those changes and updates which can make your life easier. In this edition, we announce a change to Elasticsearch replica settings and we round up last week's Compose Articles.

Elasticsearch replica tune-up

At Compose we regularly review how our database deployments are used. Our Elasticsearch team have just finished such a review and have made a number of internal changes to improved reliability and performance. Most of these will be invisible to customers, but there is one change that is going to be visible to users - we're changing the default number of replicas on Elasticsearch.

Replicas are duplicates of your dataset held on the cluster and, from our introduction of Elasticsearch and with caution in mind, we've set the default number of replicas to two. That means that for every index, there are two other copies of it on the cluster on different nodes. Although this is very resilient, we've been concerned that for many Compose Elasticsearch users this has not been the ideal configuration.

Experienced Elasticsearch users have been able to change their defaults and benefit from lowering the number of replicas to 1. This gives one copy of the indexes data, enough to survive a host being taken offline whilst delivering better handling and performance. Their experience and our team's data has lead us to conclude that, from now on, the default number of replicas will be set to one to better suit the typical workload we see on Compose.

We still recommend tuning your replica count to something appropriate to your workload. Read heavy workloads are best with a replica setting of 2, for example. We are not modifying existing indices with this change. Users are free to adjust the replica count for their existing indices either through the Update Indices Settings API or by contacting Compose support for assistance.

Compose Articles.

In the past week of Compose Articles, we showed how to connect NodeJS to Redis and RabbitMQ, showed how to save on your Redis deployments and rounded up the week's news.

That's it for this week's Noteworthy at Compose. Have a great week and we'll see you for the next Noteworthy in just over a week.


Read more articles about Compose databases - use our Curated Collections Guide for articles on each database type. If you have any feedback about this or any other Compose article, drop the Compose Articles team a line at articles@compose.com. We're happy to hear from you.

Dj Walker-Morgan
Dj Walker-Morgan was Compose's resident Content Curator, and has been both a developer and writer since Apples came in II flavors and Commodores had Pets. Love this article? Head over to Dj Walker-Morgan’s author page to keep reading.

Conquer the Data Layer

Spend your time developing apps, not managing databases.