OpenSearch 2.1 Release Highlights

OpenSearch 2.1 was recently released and here are the highlights:

• Snapshot Management: you could back up indices using Index Management before, but this only works well for time-series use-cases, like logs. For E-commerce applications or other OpenSearch clusters where indices constantly change, you want to snapshot on a schedule. Snapshot Management helps you do exactly that.
• Multi-terms aggregation. This mirrors Elasticsearch’s multi-terms aggregation, allowing you to aggregate on a composite key. For example, if you have a first_name and a last_name field, you could facet on the full name with a multi-terms aggregation. If you do this a lot, it will pay off to have a separate full_name field, but this aggregation lets you choose combinations at query time.
• Dedicated ML nodes. If you’re using the ml-commons plugin to do predictions, like when you may run out of disk, now you can do the processing on separate OpenSearch nodes. Before, data nodes had to do this work.

If you want to upgrade to 2.x, now’s a good time, because the 2.x branch has been out for a while. Here’s a quick reminder of what OpenSearch 2.0 brought to the table:

• Lucene 9.1, which has quite a lot of performance improvements and bugfixes, especially if you’re using point fields (dates, numbers, geo). It also adds support for k-NN search, which the OpenSearch k-NN plugin is planning to migrate to.
• Notification plugin. Now you can define a notification channel only once (e.g. an Email or a Slack channel) and use it in any other plugin, like Alerts or Index Management.
• Non-inclusive terminology removal. Most notably, the master nodes are now called cluster manager nodes.

If you need help with OpenSearch, you’re in the right place! Sematext offers:

• Consulting to help you with development
• Production support to help you during production fires
• Public and private training classes to teach you how to fish 🙂
• An observability SaaS to keep your OpenSearch metrics, logs and more