The Elasticsearch dev team are pleased to announce the release of elasticsearch 0.90.0 - the first stable release of Elasticsearch based on Lucene 4. You can download it here.
Besides all of the performance and memory improvements made by Lucene, version 0.90.0 comes with a host of improvements and new features specific to Elasticsearch itself. Probably the biggest improvement that our users will notice is much better memory usage when loading fielddata for faceting or sorting on a field. Fielddata uses less memory and makes it easier for the garbage collector to do its work, resulting in more stable clusters. This change alone makes it worth upgrading.
Other user-visible features include:
- a context-aware "did-you-mean?" phrase suggester
- the common-terms query, which makes stopwords redundant, improving search precision and recall
- the rescore search-phase, allowing the performant reranking of the top N results using a more expensive algorithm
- the terms filter can now retrieve a list of terms from another document, helping with social-graph queries
- sorting on multi-value numeric fields (even nested fields) using min, max, avg or sum, and geo-point fields with min and max
- a new shard allocator which balances shards within an index between all nodes in the cluster, and adds hooks for environment-aware shard allocation
On Thursday 2nd May at 9am PST (6pm CET) we will be holding a free (reg. required) live webinar explaining more about the new features in 0.90 and how you can use them. Come along and ask questions. You can register for the webinar here.
The next major release will be 1.00, which we have just finished planning at our dev week in Amsterdam. We will post about the new features we intend to include in 1.00 soon, but suffice to say, there are some very cool things planned. In the meantime, please download and use 0.90, and let us know what you think.