We heavily use elasticsearch as part of Kiln's infrastructure. We've actually found it to be an insanely useful tool: it scales extremely well, it's highly redundant, it's trivial to extend, and it is fast. We've managed to move every last bit of our search infrastructure (including source code), using over 100 million documents representing 182 GB of data, and we still get search times around 10 ms or less. It's so fast we've got an entirely new feature built around it coming out in the next few weeks. It's just that awesome.
The only downside we've hit with ES is poor documentation. In all other respects, if you need search, I can unconditionally recommend it.
I agree, ES is awesome. The poor documentation can be annoying indeed, but Shay Banon is always very helpful on the discussion groups so that's a big plus.
+1 on this, a better side bar or something on docs would be wonderful, but the tech just rules. The ease of setting up a redundant cluster is great, not to mention the just insane response speed. We switched a few page loads from mongodb to elastic search, mongo was taking ~13000 ms per page load, elastic took ~250 (2 mm records).
We store it entirely in RAM. ES stores data on disk for persistence, but prefers to keep its working set fully in RAM, and will (by design) crash by default if it runs out of memory.
Awesome news from a truly awesome developer working on fantastically awesome software.
Did I mention it is awesome?
If you haven't had a chance to play around with ES - you really should. I've never used Lucene or Solr but I picked up the basics of ES really quickly. You can roll out some really impressive search capabilities with embarrassingly little code (or knowledge).
Just this last weekend I migrated servers and syncing my new server was as easy as installing ES, joining the cluster and letting everything transfer over automatically. ElasticSearch comes as close to "magic" as possible, especially for something as sophisticated as search.
Shay works like mad on this project, it's awesome to see this for him and the rest of people working together on it. I've been on the ES mailing list for a number of months now and not a single question goes by without an involved response from Shay, helping people out every day.
Great news. I discovered ES last year and fell in love with it. Far better than solr in almost every aspect. When I saw how automatically detects nodes, distributes data, recovers from a failed node, etc. all without administration I could not believe it. It's every devops dream.
Been using elasticsearch for a number of projects and all I can say is that it's really great! We almost replaced it on one large project to use it as a distributed storage but didn't work out cause it couldn't handle the load. (not surprising since that's not what it's exactly built for..)
However, as a search solution, elasticsearch is one of the best I've used this far (used solr, sphinx and fulltext search).
I've just started getting into Elastic Search for a project I'm working on. Coming from a couple of years using Solr it is one of the best and most exciting pieces of tech I've played with in ages.
I hope this announcement doesn't change anything (at least not too much!).
The only downside we've hit with ES is poor documentation. In all other respects, if you need search, I can unconditionally recommend it.