Introducing DataStax Enterprise 4.7 and OpsCenter 5.2May 19, 2015
I’m very pleased to announce DataStax Enterprise (DSE) 4.7, a new release of our distributed enterprise database platform powered by Apache Cassandra, and OpsCenter 5.2, an exciting new version of our visual management and monitoring solution.
Our latest releases provide enhancements across nearly every major DSE component and showcase why DSE is the first and best database choice for Internet Enterprise applications. Let me give you a quick tour of all that’s new in both DSE and OpsCenter.
The Best Database for Web, Mobile, and IoT Applications
Most of you have no doubt suffered through the pain of designing an application that required different types of databases to support its functionality. You had to shard the application, aim parts of it at different OLTP, analytical, search, in-memory, etc., databases to separate the workloads, and extract-load data across all the different databases to keep things in sync.
What a pain.
Here’s the thing: nearly every Web, mobile and IoT application has this kind of ‘mixed workload’ requirement. Fortunately, DSE removes the hardships associated with building these kind of systems. With DSE, you can run OLTP, analytics, and search all on the same database cluster with (1) no competition for data or compute resources; (2) no need to ETL data back and forth, and (3) have in-memory computing capabilities thrown in as well.
Each mixed-workload component of DSE’s database platform has gotten upgrades in version 4.7
Enterprise Search Enhancements
For search, we’re very excited to introduce our unique “Live Indexing” feature that makes incoming data instantly available for search. With most search software (e.g. Elasticsearch, open source Solr, etc.) you can have a delay between when new data enters a system and when that information is available for search. With DSE’s Live Indexing, your users don’t have to cool their heels waiting for new data to be searchable.
In addition, we’ve given you the ability to tightly combine search and analytics operations in DSE. We’ve enhanced the SparkSQL query language so you can pass Apache Solr search syntax directly through a WHERE clause, which greatly extends the query capability of SparkSQL so that analytic-search queries can be run in one statement. You can also run both DSE Search and DSE Analytics in full on the same nodes in a DSE cluster.
Additionally, DSE Search now sports an auto-retry mechanism that automatically reissues search queries to another replica if the originally targeted node containing search data goes down. Add to that production certification of Solr 4.10 and you’ve got a lot of search goodness that’s gone into DSE 4.7.
The analytics functionality in DSE gets a boost in 4.7 with enhancements like built-in high availability for Spark analytics, support for machine learning, and production certification of Spark 1.2. Something else we’ve been asked for is better insight into Spark performance, and we’ve delivered with support for Spark in DSE’s Performance Service. Now you can easily collect Spark metrics and query the accumulated diagnostic information from any CQL-enabled tool to more quickly tune Spark performance.
More Memory Please
DSE’s in-memory option is not about Cassandra becoming an in-memory database, but rather it brings in-memory computing to Cassandra, which is a subtle but important distinction. The in-memory option was added to DSE because our customers were constantly marrying Cassandra to some memory-based cache like Redis to satisfy the parts of their application that needed fast in-memory read performance.
In version 4.7, the in-memory option has been enhanced to utilize both on and off Java heap memory, which extends the amount of addressable RAM for memory-based objects by over 50x. Plus the simplicity of in-memory objects stays the same: an in-memory table won’t lose data and acts as any other Cassandra table with the exception that all read operations are guaranteed to be handled in RAM.
Improved Data Protection
DSE already has a pretty powerful security feature set, but 4.7 adds extra protection via enhanced transparent data encryption. In 4.7, encryption keys can be stored on non-DSE cluster machines, providing a separation of encrypted data and the encryption keys for added safety. Also, we’ve added data-in-motion SSL-based encryption for Spark analytics.
Certified Cassandra 2.1
Last but certainly not least, DSE 4.7 includes a production-certified version of Cassandra 2.1, bringing features such as user defined types, improved counters, more efficient repair operations, and much more to DSE.
Management Anywhere and Everywhere
Accompanying DSE 4.7 is OpsCenter 5.1.2, which packs support for our latest DSE release as well as some nice feature additions. Improvements include easier monitoring of in-memory tables, the ability to invoke the Spark web interface in OpsCenter for those wanting to use it, alerts for notifying administrators when non-DSE servers housing encryption key information cannot be contacted, improved graphs for monitored statistics, and support for Cassandra 2.1 incremental repair.
We’ve also pre-announced version 5.2 of OpsCenter, which includes some exciting new additions that will be available at the end of June.
With OpsCenter 5.2, we’ve made it easier to provision multiple data centers and clouds in point-and-click fashion. In addition, it’s easier to monitor the performance of a database across DC’s and understand the latencies between each DC that is included in a database cluster.
In addition, OpsCenter 5.2 delivers a long-standing security request from customers: integration with LDAP and Active Directory (something we added to DSE in version 4.6). Both may now be used to manage OpsCenter security and control who can administer and monitor clusters.
We’ve also added new monitoring/tuning interfaces that help guide admins through better understanding performance metrics with recommendations also being provided on how to take action for issues identified in OpsCenter.
DSE 4.7 and OpsCenter 5.1.2 are available for download now, and OpsCenter 5.2 will be ready for production use by the end of June.
SHARE THIS PAGE