Spice Machines' Hadoop RDBMS Scales Real-Time Applications
|Richard Harris in Enterprise Saturday, October 31, 2015|
Splice Machine has released version 1.5 of its Hadoop RDBMS, adding multiple features that allow companies to replace their traditional RDBMSs, such as Oracle & MySQL, to power real-time applications, run operational data lakes and accelerate their ETL pipelines.
Splice Machine’s Hadoop RDBMS is designed to scale real-time applications using commodity hardware without application rewrites. The Splice Machine database is a modern, scale-out alternative to traditional RDBMSs offering a full-featured Hadoop RDBMS with ACID transactions.
Splice Machine is built on Apache Derby and HBase/Hadoop. This allows for distributed, parallelized query execution that works with all of the standard Hadoop Distributions. Features include an ANSI SQL database with the ability to scale out to petabytes of data. The Hadoop RDBMS offers real time updates with transactional integrity, distributed, parallelized query execution and high concurrency on a flexible general purpose database platform.
New features included with version 1.5 of Splice Machine’s Hadoop RDBMS include:
- SQL Compliance: Foreign Keys enables referential integrity between tables and triggers provides the ability to execute procedural code based on update, insert or delete actions.
- BI Tool Compatibility (e.g. Tableau, MicroStrategy) enables BI Tools to create temporary tables to perform operations on intermediate results. SQL Functions offer pre-built functions such as TOP N, LIMIT N, MONTHNAME, QUARTER, WEEK and NOW.
- Incremental Backup provides recovery from data corruption by backing up data changed since last backup.
- Window Functions provide FIRST, LAST, LEAD and LAG functions for advanced analytical queries.
- Enhanced Statistics collect more granular statistics required to choose best query plan.
- Cost-Based Optimizer Improvements enhance the ordering of table access, selection of query plans, index selection, join algorithm selection (e.g., broadcast join, merge join, merge sort join, batch nested loop join)
- Subquery Unrolling converts nested subqueries in joins, accelerating certain analytical queries by well over 1000 times
- ETL Acceleration abilities through Splice Machine transactional capabilities ensure that Hadoop handles ETL errors and data quality issues without reloading all of the data. Splice Machine can also enable incremental ETL that can drive down ETL lag times.
The Splice Machine database is available for licensing at a per node price.
Read more: http://www.splicemachine.com/why-splice/