Arthur Cole spoke with Rohit Kshetrapal, CEO, Tegile Systems.
Despite the best efforts of virtual desktop developers, the enterprise still has to confront some harsh realities when deploying VDI. Chief among them is the fact that as deployments scale, the cost of additional storage starts to outweigh the savings on the client side. New approaches to this equation are starting to bear fruit, however. Companies like Tegile Systems claim upwards of 75 percent cost reductions in VDI storage. CEO Rohit Kshetrapal explains how it's done.
Cole: Most enterprises show initial enthusiasm for VDI, but then balk at the storage costs. Is there any way to improve the storage situation?
Kshetrapal: We see the exact same issue. Many enterprises are interested in VDI, but at the cost per desktop that traditional storage hits, it is cost-prohibitive. If you look solely at storage, VDI simply shifts cost from the laptop to a shared storage system. The sophistication and resiliency built into shared storage systems drive cost well beyond that of a simple 2.5-inch drive in a laptop. What that means to a VDI program is that storage costs can double, triple or even worse. Sophisticated data reduction algorithms such as in-line de-duplication and compression are required to shrink the VDI footprint and reduce costs. These technologies, as implemented in Tegile's hybrid storage arrays, bring VDI storage costs down as much as 75 percent.
Cole: SSDs are often cited as an answer to VDI storage woes. But are there ways to incorporate legacy HDD infrastructure as well?
Kshetrapal: Legacy storage can be used in small VDI implementations, but at scale, they lose performance integrity very quickly. Tegile's field engineers run into this quite often. Customers give a small VDI implementation a try, but as they add users their storage performance becomes a blocker to scale. A new balance between performance and capacity must be established. Hybrid arrays do just that for VDI projects.
Cole: You mentioned in-line deduplication. Are there ways it can be used to reduce storage requirements without hampering application performance?
Kshetrapal: If implemented properly, in-line deduplication can actually boost performance. Tegile's Metadata Accelerated Storage Systems (MASS) do just that. In-line de-duplication and compression are designed into the array before the DRAM- and SSD-based cache. This allows users to put much more data into cache, thus boosting cache hit ratios dramatically. In both virtual server and desktop environments, we see performance increase as much as 7-fold compared to traditional arrays.