Last week, I participated in an interesting #ibmblu twitterchat on “Workload Management for Growing Data.” The timing of this twitterchat was quite ironic for me, as last week was one of those weeks where I constantly felt like I was never on top of anything. The big stuff kept occupying my time, so I never made any progress on the little things. And, even worse, context switching amongst all the big things meant I was making little progress. I could have benefited from some improved workload management!
Many experts participated in the lively debate about the speed of thought requirements for analytics on big data. You can view a streamlined transcript below or here.
All the twitterchat participants agreed that in today’s era of big data, low latency workloads are becoming the standard, not the exception. As Richard Lee (@InfoMgmtExec) stated, “Latency is the Enemy!”
Throwing more hardware at the problem is not a preferred approach, nor is it cost effective! As a result, growing volumes of data are putting even more pressure on analytic solutions to do more mixed workloads with the same or fewer resources. Add the mix of real-time and batch jobs and there is a stronger requirement for governing workloads. Workload management can solve both requirements.
Workload management (WLM) ensures resources are used effectively based on business priorities. It prevents servers from being overwhelmed, ensuring all hardware resources are working efficiently to deliver reports at the speed of thought. Although you’ll miss your coffee breaks from the long running queries, more work is done in less time.
With automated workload management, the hard work of defining the WLM policies is done for you to protect the system from thrashing. DB2 10.5 with BLU Acceleration not only provides automated WLM, but is effective at implementing the automated policy because BLU reduces the variance in run times, ensuring consistent response times.
Which leads us to the most interesting discussion in the twitterchat, from my point of view: Are SLAs still relevant? Many argued the point that SLAs are too static a requirement and can’t reflect the constantly changing business needs. The new expectation for service is right now, regardless of how big the data gets.
Although analytic workload demands are increasing in complexity and volume, BLU Acceleration and automated WLM offer the technology to meet the demands with a simple automated approach. BLU Acceleration provides the technology to process queries with optimal performance over growing data volumes and complexities, while automated workload management is the traffic cop that keeps traffic moving and balances the system resources across the workload. Combining the two is a win-win situation!
Join us for our next #ibmblu twitterchat on Wednesday, October 2, when we'll discuss “Striking a Balance: Disk & Memory.”