IDAA and Analytics – Changing the Face of z Systems

July 8th, 2016

J Jeffrey Broderick
z Systems Architect

The IBM DB2 Analytics Accelerator has been dramatically changing how customers view analytics, especially with the preponderance of data already existing on z Systems. In many ways, IDAA is a disruptive technology for analytics and z Systems.

Customers, in order to provide suitable performance for their analytics solutions, have historically extracted and built entirely standalone data warehouse repositories (usually referred to as OnLine Analytics Processing – OLAP). Many of these data warehouse repositories also required extensive data re-engineering—specifically built to facilitate the “type” of analytics query that the end user might request. There would be an entire infrastructure of Extract-Transform-Load (ETL) jobs and jobstreams built to move and manipulate the originating data (usually referred to as OnLine Transaction Processing – OLTP) into the new data warehousing structure.

Any changes to originating data structures would require re-building of the warehouse structures, modifications to the ETL processes, along with coordinated implementation across them. Due to processing requirements, these same ETL jobs may only be scheduled weekly or monthly resulting in “stale” data, or data that significantly lacks currency. Customers are also required to direct their analytics tooling at the data warehouse vs. directing it at the production data, even though they may have other current reporting requirements for production data. And then IDAA came along….. IDAA is a disruptive technology for z Systems because it completely changes the way we look at data for analytics.

IDAA disrupts by “converging” the concepts of OLTP and OLAP into a cohesive “data-centric” solution, that can be “accelerated” to meet the analytics needs of end users. DBA’s only need to design one version of the data structures – that which meets the needs of OLTP systems. IDAA makes that same data transparently accelerated, due to its unique architecture—without any extraneous ETL processes, major time delays or manual intervention. The data can be synchronized in near-real time to facilitate extremely high currency of information.

Physical facilities (hardware) are also substantially reduced:

  • No need for separate, duplicated storage for placement of data warehousing data
  • No need for duplication of massive servers or specialized servers just to support a data warehouse
  • No need to tolerate weekly or monthly delays in data currency
  • No need to consume extra processing cycles just to move the data somewhere else

The true simplicity is realized because end users are not required to make any modifications to their existing analytics tools (assuming they were connecting to DB2). A query against production DB2 can be seamlessly offloaded to the IDAA—and transparent to the end user.

Mainline