Thank you to all of the customers, colleagues, and IBMers who joined The Fillmore Group on April 11, 2013 at Cinghiale in Baltimore, USA to discuss IBM’s Data Strategy – which we call “DB2 101”.  As it happens, many DB2 customers don’t realize the strength and depth of the IBM Information Management portfolio.  And many non-DB2 customers don’t know about the governance and management tooling available in InfoSphere and Optim (e.g. Guardium) which IBM provides for Oracle, SQL Server, Teradata, and other-vendor database servers.

Prominently featured were the highlights of IBM’s April 3rd announcement held at Almaden Lab which my colleague, Kim May, and I attended.

• PureData System for Hadoop
• DB2 for LUW V10.5 with BLU Acceleration

Also of note is the bundling of functionality in DB2 Advanced Enterprise Server Edition (AESE).

1. pureScale – shared-disk clustering
2. Data Partitioning Feature (DPF) – shared-nothing hash partitioning
3. DB2 for LUW v10.5 columnar compression database

The presentation is available at DB2 101 – IBM Information Management Software Portfolio Overview 2013-04  The audio-video recorded delivery can be viewed here.

Announced on October 8, 2012 IBM PureData Systems were heavily featured in the Information on Demand conference opening keynote this morning.  There are three members of the PureData family:

• PureData System for Transactions (PDST)
• PureData System for Operational Analytics (PDSOA)
• PureData System for Analytics (PDSA)

I’m using the abbreviations for convenience and to save keystrokes.

PDST is built on IBM’s pureScale technology; pureScale is the shared-disk cluster technology for DB2 running on Power and Intel processors.  IBM based pureScale on the DB2 for z/OS Parallel Sysplex fault tolerance that mainframers have been using for years.  The relatively slow adoption of pureScale has been a puzzlement.  High-availability database solutions are being adopted by government and industry where the results of even a brief outage can range from bad for business to catastrophic.  One inhibitor to adoption has been removed, that of the InfiniBand requirement for cluster communications.  The PureData System for Transactions uses a simple Enternet connection for lashing together nodes in the cluster.

PDSOA is the next generation of the IBM Smart Analytics System (ISAS).  ISAS was arguably IBM’s first data “appliance” – a preconfigured, balanced processor/memory/storage stack bundled with InfoSphere Warehouse.  InfoSphere Warehouse employs the shared-nothing MPP functionality formerly known as Data Partioning Feature (DPF).  One prospective use of PDSOA would be near-real-time fraud detection.

PDSA is Netezza running under the covers.  Using innovative function-to-data Field-Programmable Gate Array (FPGA) integrated circuits, Netezza achieves orders-of-magnitude greater query performance for deep analytics like market-basket analysis.

The goal of the entire PureSystems portfolio – which includes offerings for WebSphere and custom applications – is to accelerate time-to-value and reduce or eliminate the repetitive, mundane tasks involved in infrastructure deployment.  IBM says that their PureData Systems go from door-stop to “load ready” in 4 hours.  Enterprises that provision database servers regularly commit weeks to these tasks.

One telling anecdote is that when the PureSystems plan was presented to then IBM CEO Sam Palmisano, Sam’s reply was that team was “building the AS/400”, a rock-solid platform known for its integration (hence the “i” in iSeries and System i), simplicity, and ease of maintenance and use.  Indeed, two of the PureData servers are manufactured in the AS/400 Rochester, MN IBM Lab.  PureSystems are nicknamed “AS/500”.

There’s lot’s more detail to come on the PureData varients.  Think of this as a cheat-sheet for telling the siblings apart.