Elastic 8.0 Review – First Impressions

Elastic 8.0 Review – First Impressions


Initial Thoughts on Elastic’s Latest Release

Elastic recently announced the availability of Elastic 8.0. This blog is not meant to be an exhaustive Elastic 8.0 review, but we wanted to share our initial thoughts and impressions. The official release page is text and marketing graphics heavy and meant for a general audience. We found this reference page of release highlights to be much more succinct, and this TechTarget summary is also brief and to the point.

Elastic 8.0 new feature summary

The reference page highlights what’s “new and improved” for Elastic 8.0:

  • x REST API compatibility
  • Security features are enabled and configured by default
  • Better protection for system indices
  • New kNN search API
  • Storage savings for “keyword,” “match_only_text” and “text” fields
  • Faster indexing of “geo_point,” “geo_shape,” and range fields
  • PyTorch model support for natural language processing

This is an odd assortment of highlights, so we will comment on what we think the most important developments are, with the most interesting ones first.

Improved Vector Search with kNN API

Vector Search is not new to Elastic, but as we mentioned in a recent blog, we think that it is becoming increasingly critical in advanced search applications.  Vector Search converts blocks of text and unstructured content into mathematical form for precise similarity comparisons. The incorporation of a new kNN search API helps to ensure that vector search can be applied on a large scale.

Better natural language processing with PyTorch model support

Elastic doesn’t pretend to be the end-all platform for all things machine learning. PyTorch is an open-source framework purposely designed for machine learning and deep learning models. Elastic 8.0 supports models trained outside of Elasticsearch for use at ingest time. This is helpful for complex natural language processing (NLP) applications where the model development and enhancement processes are decoupled from the application architecture.

7.x REST API compatibility to ease migration pain

While this highlight doesn’t qualify as a cool new feature, we have all known the pain of version upgrades in application maintenance. Elastic 8.0 introduces several breaking changes to the Elasticsearch REST API. So, it is comforting – at least on the surface – that there is an attempt to ensure backward compatibility for applications currently developed to use 7.x versions.

Security and performance enhancements

In our cursory Elastic 8.0 review, the remainder of the release highlights aren’t worthy of significant commentary beyond what’s already documented by Elastic.

  • Security features are enabled and configured by default
  • Better protection for system indices
  • Storage savings for “keyword,” “match_only_text” and “text” fields
  • Faster indexing of “geo_point,” “geo_shape,” and range fields

These and other more minor changes fall into the general category of security and performance enhancements, which every release tries to improve upon. Some of these enhancements are naturally related to enhancements in Apache Lucene 9.0

Is this really a “major” release?

The features to enhance Vector Search and natural language processing support with PyTorch are exciting developments and shows that Elastic value their importance in the evolution of search application architecture. Future benchmark tests may validate the claimed performance improvements.

But was this release really worthy of a major number? (The previous release was version 7.17 on January 2022). We did not see any significant enhancements in Kibana or other major parts of the Elastic platform. But we do think there are some important subtle developments.

Elastic 8.0 is increasingly cloud-friendly

Elastic 8.0 release highlights oddly do not mention two things highlighted in the availability announcement – though this text is only available if you click on “continue reading” on the announcement page.

Elastic 8.0 continues to try to be much more friendly to cloud-based applications with a new integration into Amazon S3 which speeds the ingestion content like large logs.  Despite past friction with AWS, Elastic certainly knows where customers are likely to store their data. We expect similar integrations with Google Cloud Storage and Microsoft Azure Blob as demand dictates.

Distinguishing Elastic from OpenSearch

While it may not have come up in release meetings, a subtle impact of Elastic 8.0 is that it is a clear version distinction from OpenSearch, which branched Elasticsearch and Kibana from Elastic 7.10.2.  As we have mentioned in our blog series on Elasticsearch vs OpenSearch, the latter could emerge as a viable, pure open-source alternative to Elasticsearch and Kibana.  It will be interesting to see if and how these two solutions diverge or follow similar paths in the near future.

I hope you found this brief Elastic 8.0 review and our initial impressions helpful. I think the hyperlinked content will prove useful as well and there may be performance benchmarks published in the future to verify the claimed performance improvements.

If you have any questions about Elastic 8.0, PyTorch, kNN or Search, or anything else, drop me a note at info@pureinsights.com.




Stay up to date with our latest insights!