We’ve been talking all about it, now it’s here! Download Sphinx 2.1.1-beta to see what lightning fast search is all about.
Summary
- Better real time indexes, they’re catching up with on disk indexes!
- Built in options for high availability
- Fulltext search and ranking optimizations
- JSON attributes
- Wordbreaker
- New SphinxQL commands and options
RT Indexes
Realtime indexes were lacking some of the functionality of on-disk indexes. In Sphinx 2.1.1, realtime indexes got a whole bunch of elbow-grease! Read all about it here.
High Availability
In previous versions of Sphinx, external load balancing or mirroring tools were required for building failover clusters. This came with some disadvantages which were a big deal to some in our community. But you don’t have to worry any longer! Sphinx now offers built in tools geared specifically for your clustering needs! Look for more about HA in future blog posts.
Fulltext optimizations
With Sphinx 2.1.1 you can speed up your phrase searches with bigrams, simplify complex boolean expressions, and use our BM25 function (which now includes field lengths). If you’re a frequent visitor of the Sphinx blog you’ll soon see more information on these.
Moreover, both text search and indexing generally got faster! Our UTF-8 text processing code path got heavily optimized, and that made the entire indexing up to 10..25% faster. And for text search, we implemented skiplists, so that queries that mix rare and frequent keywords (think of something like “the hobbit”) execute faster. In extreme semi-synthetic cases they execute up to 10x and even 100x faster. But general search performance also got better, for instance, on a small test collection of 1,000,000 blog posts we’re seeing a 70% performance increase (from approximately 400 qps/core to approximately 700 qps/core).
JSON attribute
Sphinx now supports JSON attributes! We’ve got some more work to do, but we’re pretty happy with our accomplishments. Learn more here.
Wordbreaker
To make indexing URLs easier we’ve added a new tool: wordbreaker. Want to learn more? We wrote about it here.
New SphinxQL commands
We’ve added some new SphinxQL commands and options. You may have already read about these new commands, but if not – this blog post describes them.
Conclusions
So, we just scratched the surface of what’s new in 2.1.1. And really – we could yammer-on about the vast array of exciting features in Sphinx 2.1.1 forever (or, at least for a very, very long time). Instead, we’d rather have you give it a try and see Sphinx in action for yourself.
Get ahead of the curve. Take full advantage of these new features sooner (and without unnecessary headache) by enlisting the support of the Sphinx team!
If you have any questions or require support - let us know!
Happy searching!
– The Sphinx Team
| « February 12, 2013. Comparing Compilers | March 8, 2013. Sphinx @ Percona Live: MySQL User Conference » |
Cool.
Please add the beta version for windows
Thanks
Thanks for this great release.
Keep the performance upgrades coming! I am, and will probably continue to be, a “regular” indexer-based sphinx user. The more features and benefits that come available to me, the more my mouth waters!
Something that I would like to see is the ability for searches to use multiple cores/threads by default (rather than setting up complex distributed indices to achieve this).
Regardless, I am still very pleased with Sphinx and thankful for your work. Keep the updates coming…
@Jonn: windows version will be added soon.
Very interested in the HA stuff!
Sphinx needs to add Cassandra support…very important to beat lucene
Are there any further information on the HA features?
@Michael: we’ll publish soon one or more posts about HA features.
Will there be a 32bit version for windows? Thank you for this great release. :)
@adrian when for Windows avaliable, much better if avaliable for 32 bits or 64 bits too :-)
@e-Padi Windows 32&64bits builds already available :)