So much has been learned and written about how to find and fix storage related performance problems that I wondered if there’s anything new that can be done. There are tons of best practices, monitoring products, and professional services all dedicated to solving this often quixotic burden. But at the end of the day, after all the research, analysis, meetings, escalations, finger-pointing, begging, threats, firings, and money thrown at it, there remains one final act that throws more terror into the hearts of men and women than anything else.
But before we go there, let’s briefly recall the trip to the solution. It usually starts when an application user complains of an abnormally slow response time. IT mobilizes to uncover the problem and the search for the guilty begins. Everyone then proceeds to try to prove the innocence of their domain. Vendors are brought in to prove that it’s someone else’s fault. Meetings are held, fingers are pointed, reports are reviewed, but eventually, a solution is suggested. It could be a configuration change, a hardware upgrade, all new gear, or even new application software. The decision is made, the money is spent, and everyone holds their breaths and crosses their fingers as the change goes “live” in production. The moment of terror.
We’ve all been there. Sometimes the problem goes away, sometimes it doesn’t, and sometimes, the change makes it worse. More often than not, the whole process is repeated. So what’s the missing link? There’s no way to know if the change is actually the fix. Or is there?
What if you could replicate the problem workload in a lab environment, allowing you to test the change(s) in a controlled, pre-production manner? Well, you can. You can employ the Load DynamiX Workload Sensor to capture the problem application workload, and replay it in the lab with the Load DynamiX Workload Generation Appliance. You can test your suspected fixes until you actually find the real solution.
Why hasn’t everyone been doing this all along? They couldn’t. Mostly, it’s because the synthetic workload models weren’t accurate enough for troubleshooting. They were good enough for a lot of other purposes, but they weren’t based on incredibly granular, actual production I/O – essentially a real-time capture of your production workload profiles. What’s new is the Load DynamiX Workload Sensor, just announced and shipped in December 2015. The Workload Sensor is used to build the application workload model from the actual production workload. No guesswork. No terror.
To learn more, click here, or better yet, talk with a Load DynamiX representative and ask to see a demo.