You are here

NuoDB Robins Release 2.3 is here

I closed out the blog post announcing our last release, NuoDB Larks Release 2.2, like this:

See you in three months, when NuoDB Robins hits the streets. We think you'll like it :)

We've been busy with Robins aka Release 2.3. We think Robins is important for a number of reasons.

First, as with all of our releases, Robins includes hundreds of specific improvements based on customer scenarios, representing heavy investments in all aspects of our infrastructure. Robins, in general, is the fastest, most reliable, and easiest to deploy and operate version of NuoDB we've shipped.

Next: Robins introduces a preview version of "Continuous Data Protection" (CDP), which allows operators to continuously and non-invasively snapshot all transactions in a running database. Unlike with a typical backup, you can easily restore a database up to a specific transaction, which is critical in compliance or "fat fingered catastrophe" scenarios. 

Another preview capability, Table Partitions, adds the ability to specify a partition policy on a table (such as RANGE or LIST) so that rows meeting a criteria are logically mapped to specific table partitions; this enables important scenarios such as aging out of historical data, or improving the performance of certain kinds of queries. A related preview capability, Storage Groups, takes table partitioning to the next level by exploiting NuoDB's unique ability to service a database with multiple Storage Managers. Specifically, an operator can map each table partition to a a Storage Group, which can then be mapped to one or more Storage Managers. The end result is that, at the operational level, you have control over where versions of each row in the database are stored. This is a huge advance in operational databases, enabling Residency scenarios and offering an important tool for performance and scalability.

Both of these preview features will be supported in production in a future release (to be announced later in the Fall), and help us plot a release trajectory established over the past year to reinforce NuoDB's ability to play a central role in an enterprise's data architecture.

Also, with Robins we've worked closely with customers to certify new geo-distributed database scenarios, expanding the Disaster Recovery support introduced in Larks with coverage for more configurations and scenarios, and also adding support for specific "active/active" scenarios for managed and unmanaged databases. This support shows in the form of an operations manual detailing use cases (such as inter- and intra-data center network partition, or loss of a specific host or container) and what to do about it, a suite of operational scripts, and several new database management commands. The result is a simpler and more reliable approach to managing a geo-distributed database in failure conditions while maintaining overall availability.

Performance

Finally: Robins is about performance. We've had a specific focus on performance with Robins, with targeted improvements in the infrastructure of Transaction Engines (TEs) and Storage Managers (SMs), and the SQL layer responsible for efficient parsing and optimization of queries. We've improved performance by more than 2x in certain write-intensive scenarios. We've improved the performance for many scenarios in which a second Storage Manager is started and synchronizes from a running Storage Manager, with a new "fingerprinting" algorithm, with dramatic improvements in both "hot" (when the second SM is starting from an existing archive) and "cold" cases. And we've also improved the performance of certain SQL queries involving GROUP BY and ORDER BY, for queries involving a single table and no parameters.

Most importantly, Robins previews a new set of capabilities, Table Partitions and Storage Groups. As noted above, these features work together to activate key business or logical-level scenarios. But we're especially excited about how they open up a new set of outcomes when it comes to performance and scalability at the operational level. The abiity to specify where data of various kinds is stored means that operators have unprecendented flexibility to shape disk and network loads within and between data centers.

A summary of the other new features and capabilities of this release:

  • Admin and Management Layer: working closely with early adopters, we've added features, utilities, and certified procedures for handling important intra- and inter-data center failure scenarios, such as: loss of a single or multiple TE, SM, broker, or host, including more than half the hosts in a domain, due to failure or network partition. The improvement here is that in some of these scenarios, the affected database would shutdown due to the lack of a majority, but with these added capabilities, the operator has new options to repair the domain (and form a new majority) from the surviving hosts.
  • The NuoDB manager command line utility, nuodbmgr, has been enhanced with new command line input handling features, via the JLine 2.x utility, providing functionality similar to the interactive input and editing capabilities available in modern Unix/Linux shells (such as bash and tcsh).
  • SQL support for updatable views, where the VIEW contains a reference to a single base table
  • Java 8 support: Completing the transition started with NuoDB Swifts Release 2.1, we've now added support for Java 1.7 and Java 1.8, and removed support for Java 1.6 for NuoDB hosts. 
  • Client applications are unaffected. See the documentation for more info.

In Closing

For more information about Robins, please visit the release notes.

As always, it's important to acknowledge the critical role our customers and partners play in helping us make the NuoDB distributed database a better product!

Robins is available as a Release Candidate to customers and qualified prospects.  Look for general availabiity of Robins in the next two weeks at www.nuodb.com/download.  Please let us know what you think.

(And yes, see you in three months with news about NuoDB Cranes... note that we've actually been working on Cranes and it's follow-on companion release Finches for a while now, in parallel with Larks and Robins. Cranes and Finches will deliver further and more deeper support for geo-distributed database scenarios as we continue to execute on our journey towards generalized support for global operations.)

Add new comment