As the prices of solid state drives (SSDs) fall and new choices emerge on how the storage resources can be used better, administrators are thinking of ways to tap technology to enhance storage performance.
It is also a fact that many data center administrators and managers have seen performance improvements when they used SSDs on computing devices of end-users. The problem, however, is that when they try to improve a data center’s storage performance, it becomes quite complex.
If the hypervisor is used to its potential with the new feature of vSphere Flash Read Cache that is available on vSphere 5.5, results do improve on the storage performance front. Likewise, in the VMware offering, the newly introduced VSAN feature may be another alternative that can be tried out. VSAN, which is akin to the Flash Read Cache, offers a caching solution across nodes to bring out from local storage systems a logical storage capacity. SSD is needed to an extent in VSAN and doubles up as a shared storage disk structure for VMware VMs.
VSAN, a hyper-converged infrastructure technology, merges CPU/Memory of servers with the role of shared storage. Windows Server 2012 R2 provides a tiering solution with Storage Spaces on the Hyper-V side.
Besides the VMware solutions, a suite of storage systems now optimally use SSDs. Most of these storage systems are a combination of SSDs and conventional rotating storage, where features and technology variation will stretch workload performance by tapping SSDs fully where required.
It is essential to comprehend how these specific deployments vary. Each of them offers benefits that are different from one another. It is the job of the data center administrators to determine how much money is to be spent for the variations in performance improvements.
There are certain things which we should avoid doing. For instance, we should not insert SSDs without taking quite a few factors into consideration. There are different types of SSDs: Consumer and Enterprise-class. The data center deployments should, however, adhere more with the enterprise-class patterns. Disk controller systems should also be factored into the decision-making process. SSDs from a bus and interface standpoint may be maintained, but disks could be used improperly. Improper performance could cause irregular wear, hastened failure, and investment would be squandered.
There are ways we can go about rectifying these blind spots. Firstly, understand what the SSDs will benefit. If it is the virtual infrastructure, make available a high-performance disk tier to VMs makes owing to the amalgamated and handy nature of VMware and Hyper-V environs.
The next point to be considered is if a native hypervisor solution with the SSD advantages is needed or a storage system. It would be an easier procedure to change all hosts to put in place an enterprise-class SSD and tap technologies, such as VSAN or vSphere Flash Read Cache. If you opt for the storage system, it would incur a higher investment, but it lets true tiering vs. only caching to be introduced.