The Clustrix team is proud to announce 4.1 release
Some of the key highlights from 4.1 include the following:
Fast Parallel Backup
Clustrix now implements fast backup and restore as a binary backup mechanism that works at the row level. Each Clustrix node sends its data directly to the backup target in parallel, eliminating bottlenecks and allowing backup to scale linearly with the number of nodes in the cluster. Similarly, for restore, the initiating node coordinates with other participating nodes in parallel to read from the dump file and restore replicas.
Clustrix supports mysqldump, but with Clustrix parallel backup, you can see a time savings of 20% plus additional performance gains now that Backup is now fully parallelized across all nodes. For a three-node cluster (the minimum node configuration for Clustrix), you can now expect backup to be four times faster.
Support for Fast Temporary Tables
Clustrix now provides support for temporary tables, as described in the MySQL Specification for Temp Tables. These can be useful in optimizing your application development for performance.
Internal Improvements to Optimize Performance
There are multiple performance optimizations in this release. One of our biggest is Query Fanout, which which enables large (OLAP-style) queries to utilize multiple CPUs on each node which it accesses. This increased parallelism delivers results faster for complex, data-intensive queries. The Clustrix query compiler automatically partitions the query into concurrent fragments to leverage this capability without any extra effort by the query writer.
In addition to Query Fanout, we made optimizations to how BI (Business Intelligence) queries that make use of aggregation functions (min(), max(), sum(), count(), any query with a GROUP BY, etc) are processed in parallel. These aggregates are now processed in a more distributed way, rather than performing the aggregation on a single node. This functionality is currently in beta, but can be enabled on your cluster with some coordination with the Clustrix support team. Distributed aggregates will be generally available in the upcoming months.
These performance enhancements underscore Clustrix’s ability to process OLTP and OLAP queries simultaneously. They also add to the ways Clustrix applies principles of distributed computing to scaling your OLTP workload while providing real-time analytics.