We are pleased to announce Volt Active Data v8.1. This release includes features that can reduce the total cost of ownership of your Volt Active Data solution, with improvements to memory management of replicated tables, and performance improvements for many applications. These features include:
- Reduced Memory for Replicated Tables
Replicated (i.e. non-partitioned) tables are now stored only once per cluster node, rather than being stored once per partition. For replicated tables with a large number of rows or on clusters with large sites-per-host, this can save a significant amount of memory and reduce the total cost of ownership for your Volt Active Data cluster.
- Faster Bulk-loading of Replicated Tables
If you’re using one of Volt Active Data’s loaders, such as csvloader or jdbloader, or have written your own loader using the BulkLoader API, you may notice up to a 3x speedup for loading replicated tables.
- Improved Ad Hoc Read Query Performance on Replicated Tables — Ad hoc queries that perform read-only operations on replicated tables have traditionally been executed by a single partition within the database, assuming such queries are a small percentage of the overall workload. V8.1 changes the execution model to “round robin” read-only queries of replicated tables to more evenly distribute the workload.
- Performance Optimization of “Paging” Queries
Previously, using SQL LIMIT/OFFSET to get to a range of values would have to scan the table to determine the starting point, causing each successive “page” to take longer to get to. With this optimization, LIMIT/OFFSET queries can take advantage of an index on the table to directly access the data.