Avoiding the 'I/O Blender' Effect
Condusiv takes aim at Windows I/O requests to optimize virtual workloads.
- By Dan Kusnetzky
Recently, I had an interesting conversation with Brian Morin, SVP of Global Marketing at Condusiv Technologies. At first, I thought I was speaking with a representative of a newcomer to the market for I/O and storage optimization. During the discussion, I learned that I had spoken with it many times before, when the company was named Diskeeper.
Solving the Windows Problem
The conversation centered on what can be done to improve the performance of virtual workloads hosted on Windows. The challenge is that the Windows I/O subsystem has been optimized to provide pretty good performance for all types of workloads, but virtual workloads use systems differently. The net effect is that Windows, in the attempt to respond to all the I/O activity of many virtual workloads, has a tendency to write to storage using frequent, small, choppy requests. This often results in unexpectedly slow I/O and a reduction in how many virtual workloads a given system can support. The folks at Condusiv call this "the I/O blender" effect.
Condusiv has developed V-locity, an I/O optimization product that reaches down into the Windows I/O subsystem, intelligently analyzes I/O requests (both read and write), and optimizes how storage is used. While caching is an important part of what this software does, it goes beyond that to optimize storage I/O. This, the company claims, can improve overall throughput by 50 percent or more. It supports ESX/ESXi and Hyper-V.
Dan's Take: Speeding up Reads and Writes
As I was listening to the briefing, I couldn't help but think how this technology could be useful to enterprises that field both read- and write-intensive workloads. It would, in all likelihood, be an outstanding addition to batch workloads that process just about anything from billing, payroll, health care, insurance, and/or financial information. It would provide accurate data protection and data continuity software, as well.
During my software engineer phase, I helped implement database software and related applications. I learned that these applications often faced several bottlenecks that reduced overall performance. If these applications used the network, the network was the first bottleneck to consider. Storage bottlenecks were the next in line. Memory and, finally, processing were also challenges. The Condusiv technology appears designed to address a common bottleneck.
If your company is trying to improve virtual workload performance, but doesn't have the expertise or the time to dig into operational logs produced by the applications, database, virtual machine software or the OS, it would be worth evaluating storage virtualization and optimization technology such as that offered by Condusiv.
Daniel Kusnetzky, a reformed software engineer and product manager, founded Kusnetzky Group LLC in 2006. He's literally written the book on virtualization and often comments on cloud computing, mobility and systems software. He has been a business unit manager at a hardware company and head of corporate marketing and strategy at a software company.