Times have changed. Data centers must refresh storage more frequently than ever. The organization wants faster application response to more sophisticated queries across more users. Or it wants to start a totally new endeavor, like better decision making driven by an analytics application like Splunk, Spark and Hadoop.
A storage refresh used to be an event that occurred every three years or so, where data center personnel dropped everything to start researching and testing a variety of storage systems. The goal was to find one that fit their needs and budget.
Sometimes the process was driven by a need for more capacity, but more often it was driven by time. The storage system was coming off of warranty and the cost of out-year maintenance was more expensive than getting a new system.
Given the new requirements and the increased frequency of storage refreshes, the organization can no longer expect IT to drop everything to start another refresh project. Instead, the organization needs to create a storage refresh process that operates continuously and storage refreshes need to give way to a storage performance management process.
A storage performance management process is similar to storage management. Storage management requires an IT person or team watch storage capacity utilization and predict when capacity will run out. Performance management is similar except for the fact that performance is more ephemeral than physical capacity consumption, which brings the word “process” into play.
A Storage Performance Management Process
The first step to create a performance management process is to understand what the current performance demand is and how capable the current storage infrastructure is to support that demand. Part of this step is also understanding the limits of the process. Assuming that expectations on performance will continue to increase, then the responsibility of the performance management process is to understand at what level the storage system is not able to meet performance expectations. It turns something ephemeral into something more concrete that can be used as a basis for planning.
Another responsibility of the performance management process is to try to predict when those expectations, given past requests, will occur and set a date so IT will know what parts of the storage architecture will need upgrading or replacement.
The next step in the process is to capture these workload metrics and replay them on potential new storage solutions. Companies like Virtual Instruments’ Load Dynamix family of solutions can do this in an automated fashion. The testing of new storage hardware and software should be ongoing, even if the storage system is not needed at just that moment in time. A continuous testing and change validation process provides IT breathing room to make an educated decision about which storage system or upgrade should occur next.
Once the new storage system is in place the first step is repeated on a regular basis so IT can establish new limits of the environment. These are really just the first steps in the process of developing a performance management practice. In our webinar “5 Steps To The Perfect Storage Refresh” we go in depth on the subject of storage refresh and work through the steps required to develop a complete storage performance management process.