Скачать или смотреть VMworld 2012 Session STO1430: Tracking Down Storage Performance Issues: A Customerís Perspective

With a VMware vSphereÆ implementation in several sites across four countries and three continents, Christie was well underway with its virtualization journey. The company had selected NetApp as its storage solution and had successfully grown the infrastructure to over 500 virtual machines, with many more planned for upcoming projects. However, as the numbers of virtual machines grew in the main office in Canada, strange errors started to appear. At first they were rare and very intermittent, but then began to grown at an alarming rate of frequency and intensity. Disk latencies skyrocketed to thousands of milliseconds, multiple applications unexpectedly failed over to recovery sites or crashed altogether, in-process transactions were lost, and the stability of both the storage and vSphere was called into question. Several projects were temporarily suspended until the storage issues could be resolved, including the VMware vCenterô Site Recovery Managerô implementation and storage-array virtual machine backup projects. Tracking down the problem, however, proved to be difficult. Initial troubleshooting efforts were unsuccessful in pinpointing the issue, and even falsely flagged some of the symptoms as being the root problem. Leveraging expertise from NetApp, and using different packages from third-party software venders, we gradually identified and correct issues. Every issue we identified and rectified gave us further insight into addressing the remaining issues. In the end, we discovered that it was no one of the several ìsuboptimalî configurations that were responsible for the performance issues but rather their collective influence, which placed significant stress on the storage subsystem.

Speakers:
Keith Aasen, NetApp
Scott Elliott, Christie Digital