A .Solr4.x new features
1. The near real-time search
Solr near real-time search [Near Real-Time, NRT] feature enables documents to be added quickly search, search data in response to rapid change.
2. atomic update and optimistic concurrency
Atomic update feature allows client applications on an existing document to add, update, and delete fields and other value-added operations, without the need to re-send the entire document. When there are two simultaneous requests changes to the same document, Solr using the optimistic mechanism to prevent incompatible update. Solr is simply the use of special _version_ version field to ensure the security update semantics of the document. After two requests before submitting the request to change will receive an outdated version of [this version is none of the previous two requests to perform, then the request will be executed and to submit a modified version number], it will fail to perform [request you need to verify the version, the same version can only perform the requested].
3. GET functions in real time
Regardless of whether the document has been submitted to the index, using real-time GET functions can use the unique identifier to retrieve the latest version of the [index] transaction log to provide support. This row key using the [OK] key Cassandra retrieved data key - the value stored in a similar manner. Before Solr4, unless the document is submitted to the Lucene index, otherwise it is not retrieved out. And submit a very time-consuming, affect query performance.
4. Use the transaction log to achieve sustained write
When a document is sent to Solr index, it will be written to the transaction log to prevent data loss caused by server failure. Solr transaction log is in between the client application and Lucene indexes, provide the necessary support for real-GET function, so that it can not consider whether the document has been submitted to Lucene, retrieve documents directly by a unique identifier. Solr transaction log update unbound between visibility and updates persistence. In other words, the document can be persisted, but does not appear in the search results. The transaction log can control the timing of submission of documents appearing in the search results, before submitting to avoid because of server failure and loss of data.
5. Zookeeper achieve replication and simple fragmentation
Before Solr4.x, require manual expansion. After Solr4.x, SolrCloud make expansion easy and automated. In Solr in, Zookeeper responsible for specifying the copy on behalf of fragmentation and fragmentation, and service request tracking the movement server available. SolrCloud and Zookeeper are bound, therefore SolrCloud start without making any additional configuration and installation.
To be continued. . .