DB2 pureScale at TEC 2010

I am in Austin this week at an IBM briefing and there has been a long parade of speakers, products and technologies. But, every once in a while, a topic really catches your attention.

I’ve always been keen about the IT debate whether to “scale up” or “scale out”.  In other words, do you  resolve application performance problems by adding more nodes in a cluster (scale out) or by adding more resources to an existing system (scale up)? System i customers typically and easily solve problems by scaling a system up:  adding more processors, memory and disks to an existing system. AIX and Linux customer have more of a dilemma since clustering is more commonplace and so they have options. Some applications, such as WebSphere Application Server, better lend themselves to clustering because of products such as WebSphere ND (Network Deploy) where you can string hundreds of servers together to power huge websites. Customers often view scale out as “cheaper” since they can leverage commodity processors and avoid tiered SW charges.

Database has traditionally been the holdout to the “scale out” model. Conventional wisdom is that to keep pathlengths down and reliability up, database is best kept on a single server and the larger the DB, the bigger the server. This is the typical System i response. Oracle first challenged this with their popular RAC (Real Application Cluster) technology. Oracle is aggressively pushing and “upgrading” their customer base to RAC and receives enormous attention and focus by the IT community.

What is IBM’s response? Is IBM ready to cluster DB2 as well? 

Instead of announcing an also-ran product, IBM decided to leapfrog Oracle RAC by digging deep into their enterprise portfolio of innovation, patents and products. Leveraging technology from mainframes (Parallel Sysplex) and AIX (GPFS – General Parallel File System,) IBM now has DB2 pureScale… a cluster-based shared disk architecture.

DB2 pureScale scales capacity linearly for transactional workload. It’s done by connecting a new node and issuing two simple commands. You don’t need to change your application code to efficiently run on multiple nodes. DB2 pureScale provides continuous availability through the use of IBM PowerHA pureScale technology on IBM Power systems and a redundant architecture. The system recovers nearly instantaneously from node failures, immediately redistributing the workload to surviving nodes.

It turns out the IBM Toronto Lab in instrumental in delivering this. So when I met Paul Awad, the DB2 Product Manager from the Toronto Lab, I asked if he was willing to share more information about DB2 and PureScale with the TUG and TEC crowd at our TEC2010 and he was happy to do so.

So now we’ve added an excellent session to our TEC agenda on DB2 pureScale… straight from the lab. I can imagine that this topic will be of general interest as it deals with database scalability, clustering , resilience and anyone who even remotely is interested in Oracle RAC.

Check out our agenda on the TUG website!


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: