Ask HN: Elasticsearch vs. SOLR? | Hacker News Reader

Ask HN: Elasticsearch vs. SOLR?

3 pointsAznHisoka8 comments

Has anyone got any experience with both of these? Which one would you recommend?

I have a new project where I want to store logs and search those logs by keyword, and eventually do visualizations on the performance of my app based on these logs.

View on HN

I have been involved with some complex SOLR customizations such as Asian languages, probabilistic IR and neural ranking functions, etc. I have been proud of the end product but at the cost of blood, sweat and tears. To be fair, a lot of this was in the Lucene 3 era, in which doing anything interesting with the ranking function would be a big pain and probably wreck performance.

I haven't done anything that advanced with Elasticsearch but the #1 theme is that Elasticsearch is really easy in terms of development and operations. I know people that have developed plugins for ES that are pretty clean so I think advanced customization is easier than it used to be, but a lot of that is a matter of Lucene 4.

If you are working with logs and interested in visualization you should check out Kibana and also Kibi

https://siren.solutions/kibi/

which is a greatly powered-up fork of Kibana.

We moved everything from SOLR to Elasticsearch about a year ago. We were already using Elasticsearch for data visualizations, but SOLR was our text search. Elasticsearch supported that, and we were able to add more language analyzers to support our use-case.

On the admin side, SOLR was a black box to everyone but one person on the team. If a node dies, there needs be another election, and Zookeeper needs to be updated. Restart nodes too fast and we ended up masterless a few times. Elasticsearch is much much better on leader elections and backups, though status going to red on new index creation is obnoxious. That supposedly will be fixed soon.

Perhaps not the best question for HN. But I used both a year ago and did not see a single thing in SOLR that ES didn't do better. Starting from documentation to ease of configuration and querying, features... Documentation alone was worth the switch. Overall it seems to me that SOLR is a technology that's on it's way out.

Thanks, it seems to me that Elasticsearch has better documentation as well.

Really good write-up (about a year old) on precisely that subject, here:

https://sematext.com/blog/2015/01/30/solr-elasticsearch-comp...

If your requirement is mainly indexing and search, then the performance of Sphinx (http://sphinxsearch.com) will be must better, even for large amounts of data.

I'm looking at replacing a few GSA appliances with either of these for indexing private user content (so users can search their own data). Anyone have experience with this on these and how has it been?

Unless you like XML and convoluted documentation, go for ES.

Reply on news.ycombinator.com