OpenSearch – Amazon forks Elasticsearch and the divergence begins

As expected, Amazon have announced their fork of the Elasticsearch codebase following the recent changes to the licensing of the latter by Elastic. This sounds like a herculean effort, unpicking dependencies, removing some ‘phone home’ features (which I was a little surprised to hear about!) and a lot of search-and-replace. Although Amazon are keeping the management and governance of the project to themselves there are various partners involved including RedHat, SAP and, which is an encouraging sign for the ongoing community around this project. The name is OpenSearch, and yes there was a previous use of that name.

What does this mean for users? Well, it’s still an alpha version, so you’ll have to wait a little longer before betting the farm on this fork, no matter how enthusiastic you are about its open source nature. At the moment you can still download and use a reasonably recent, open source version of Elasticsearch from Elastic themselves. However at some point you’re going to be faced with a choice and a few tough questions to answer:

  • is the product or service you’re building close enough to one of Elastic’s offerings that there’s a risk, even small, that their lawyers might enforce the (relatively untested) conditions of their new license? Remember that this licensing change came about because of competition between Amazon and Elastics over hosted versions of Elasticsearch. but there’s no guarantee that Elastic won’t consider other competitors
  • are you going to be able to rely on Amazon and its partners to build a healthy, thriving community around OpenSearch? As I’ve written before this may be key to the success of the project – maybe you can also be part of that community, a contributer not just a downstream user?
  • will there be compatible training, services, plugins and all the other things your team might need to build and augment their OpenSearch skills and capabilities, like there are for Elasticsearch?
  • will any new features developed for Elasticsearch be so important to you that you must stay on Elastic’s roadmap (even in the last week a new feature has appeared – ‘runtime fields’ and you can be sure Elastic will keep reminding you of these differences). Of course, perhaps the new features Amazon will develop like their command line interface are more useful to you. In summary, there’s going to be divergence and we don’t yet know how far this will go.

At OSC we’re already considering how we can support our clients and wider community as OpenSearch is adopted. As there’s not yet been significant divergence, our Elasticsearch training courses, relevance testing tools Quepid and Splainer, Elasticsearch Learning to Rank plugin and indeed pretty much anything built for Elasticsearch should translate easily to OpenSearch.

We’ll keep an eye on the situation, keep you updated and do of course join us in Relevance Slack or contact us directly if you want to discuss the options.

Image from Lines Vectors by Vecteezy