The most popular and comprehensive Open Source ECM platform
Let us know more
Big Data Analytics: Big Blue's BLU Acceleration
A new package it calls BLU Acceleration was recently announced by IBM Research. BLU is intended to speed up Big Data Analytics (BDA) and reporting. BLU is currently a technology announcement and not a product, but it undoubtedly being announced partially in responce to SAP’s HANA technology. BLU is based on DB2 — DB2 was extended to add a column store which can significantly speed the accessing and processing of data.
IBM has combined a number of analytic acceleration technologies and tricks into the single BLU Big Data Analytics offering. The focus of the technology was on performance. The idea is that the faster information can be processed, the more interactive and better users will be able to make decisions from their data.
- BLU uses “data skipping” which identifies and ignores information that’s duplicated or not relevant to the current processing task.
- Processing on BLU is performed in parallel across multiple processors.
- BLU can perform some analysis without first setting up a data model.
- BLU also uses “actionable compression.” With this technique BLU is able to analyze some data without first decompressing it.
- IBM’s BLU Acceleration supports the use of RAM instead of hard disks to help boost performance. It isn’t dependent on the amount of RAM available; the size of data sets could exceed the amount of available memory without problem.
When all of the combined acceleration techniques available to BLU were used, IBM benchmarking estimates that processing could be as much as 1000 times faster, with typical reporting times being sped up by a factor of 25. BLU doesn’t use indexes or aggregates, and it requires no tuning or changes to SQL.
