Since the last blog post announcing the release of YCSB 0.6.0 in Cloudera Labs, users of Cloudera CDH and EDH will have noticed regular updates to the Labs version, keeping it in lockstep with the upstream release. This should help assure users of a consistent and easy mechanism to deploy the current version of YCSB (which at the moment is v.0.10.0 in CLABS) to evaluate the performance of the NoSQL stores employed within their clusters such as HBase, Kudu and Accumulo.
There have been several updates and improvements made to YCSB since version 0.6.0, thanks to the thriving and burgeoning community. At the top of the list of salient new features are several bindings that have now been added to YCSB, helping cement its position as the performance analysis framework of choice for NoSQL data stores.
- Apache Solr users will be happy to note that there is now a new YCSB binding for Solr.
- Likewise, given the rapid adoption of S3, AWS users now have access to a S3 binding.
- The current release of YCSB includes a binding for Google Cloud Datastore, which is part of the Google CloudPlatform suite.
- Google’s CloudPlatform also includes Google Cloud Bigtable, which is the latest incarnation of the same database that HBase is based upon. YCSB has added a binding for Bigtable as well. It should be noted that Bigtable can be accessed via the HBase client too, using the hbase10 binding.
- Riak KV is yet another distributed NoSQL datastore that users can now evaluate with a newly added YCSB binding.
- AsynchHBase, the alternative asynchronous HBase client now gets its own binding.
- Yet another new binding is the one for Apache Geode (incubating), a store that prioritizes low-latency performance in conjunction with a database-like consistency model.
Enhancements and Improvements
A few of the noteworthy enhancements include:
- New POSIX-compliant launching scripts have been added. This includes support for Windows as well.
- Operation tracing support via Apache HTrace has been added. As more and more components in the stack begin supporting HTrace, this support will be helpful in identifying bottlenecks.
- JVM and garbage collection statistics are now available in the YCSB report.
- Support has been added for secure-mode HBase clusters.
- The DynamoDB binding has been updated to use the new API.
- The scan operation has been implemented for Couchbase.
- Several bindings have been updated to provide support for newer releases of the respective datastores.
In addition, there have been a large number of bug fixes and improvements, in conjunction with code cleanup. Full details are available in the release notes, which are associated with each incremental release.