Triton Db2 Geek

Confessions of a DB2 geek

IBM Gold Consultant Program and IBM Premier business Partner

DB2 LUW gains more Mainframe-style functionality

January 27th, 2010 - by

 

When you think of a database you think of rows and columns, or that is what was thought of a few years ago.  The modern DB2 data server for Linux, UNIX, and Windows (LUW) is a lot more than just a bunch of rows and columns.  It is now a powerful server able to replicate many of the features provided by the mainframe.  As each iteration of the LUW product is developed, further features are incorporated.  In hindsight this has been a long time coming, since the mainframe is recognized industry wide as the leader in data servers.

 

Today we are looking at the Workload Manager.  Why do we need the Workload Manager?  To a degree, the internet has a lot to do with the answer to that question.  As the internet has developed over the years, so have business processes.  It is not only the larger organizations developing an internet business model but there are countless smaller organizations and start-ups that are making use of this opportunity to grow.  The internet has allowed even the smaller organizations to grow into diverse markets around the world, which previously had a far greater complexity.  As these new business models have developed, so the data server has had to develop and now is required to be available 24 hours a day and 7 days a week.  Rarely are DB2 data servers taken offline for maintenance work, everything is online…9 to 5 has become 24 x 7.

 

So what does that mean for DB2? It means that the database is constantly under the strain to perform at the top of its abilities.  As more and more workloads enter the database, the ability to provide the resource to these workloads is a difficult task but not one that DB2 shirks.  The mainframe has long been able to cope with huge workloads by using its dedicated workload manager feature.  This has now been incorporated into the LUW suite of features, allowing for the manageability of diverse workloads that meet a diverse set of goals.

 

Download the WLM Technical Briefing Paper

 

The Future of WLM
With each passing release of the DB2 product, the Workload Manager also faces the prospect of a similar upheaval, with the introduction of new concepts to enhance the products’ functionality.  A new feature introduced in DB2 9.7 Workload Manager has the ability to work at close quarters with the Linux Workload Manager.  This works much in the same way as AIX Workload Manager, allowing for more control over CPU allocation to service classes.  There is a one to one relationship between the DB2 Workload Manager and the AIX Workload Manager at the service class level.

 

The integration of DB2 workload management and operating system workload management will more than likely be an ongoing integration within future releases and other operating systems.  A variety of new features have been included covering wildcard usage with workload definitions, usage of the address connection attribute with workload definitions and bufferpool priority settings when defining a service class, amongst others.

 

It will be interesting to see how the DB2 workload manager integrates with DB2 pureScale which is another of the mainframe concepts implemented on the LUW platform.  DB2 pureScale is based on the mainframe concept of data sharing.  On LUW it will aim to achieve horizontal scalability, allowing for a supply of new nodes to the existing cluster, through application transparency, no foreseeable code changes and importantly continuous data availability.  If a node fails, the remaining nodes will take over and the system will continue operation.

With an increased number of interconnected DB2 servers/nodes, what will the repercussions be for the workload manager?  A lot of these questions will in time be answered but we can speculate that, with the increased capacity the WLM has at it’s disposal several of the servers can take on the workload.  This is based on the assumption that each DB2 installation will have it’s own implementation of the workload manager.

At the basic level we could assume that a system with many workloads can distribute to the different servers in the DB2 pureScale cluster and thus workloads will be balanced amongst the servers.  This also suggests that if a server in the cluster goes down then the workloads can be distributed to other servers and can still complete.

 

We will just have to wait and see……

 

Download the WLM Technical Briefing Paper

 

Comments

« »

Tag Archives