This is a post on DCIG’s site that Ken Clipperton was kind enough to permit Tegile to post here as well.
The many business benefits of virtualizing servers include cost savings through consolidation, improved uptime/availability, improved disaster recovery/business continuity capabilities, and faster development/testing/deployment of application enhancements. In light of these benefits, virtualizing critical applications would seem to be an obvious thing to do. Yet many still hesitate because of the risk that applications will suffer performance problems and that these problems will be more difficult to troubleshoot and resolve.
VM-aware storage reduces the risk of virtualizing critical applications by restoring visibility into storage performance and in some cases by enabling performance management on a per-VM level. In the opening keynote session at VM World 2012, VMware CEOs Pat Gelsinger stated his expectation that “within 4 years more than 90% of all workloads will be virtualized.” To reach that level of virtualization, even business-critical applications will have to be virtualized. Later in the same keynote session Steve Herrod, VMware CTO, cited examples of businesses that have recently virtualized critical applications and then he highlighted VM-aware storage as a key enabler of this progress.
The risk of storage performance issues is real. As a data center moves from a “one application, one server” model with direct-attached storage to a virtualized shared server model, storage also becomes a shared resource–whether continuing with direct attached storage or through implementing a storage area network (SAN). As I have previously noted, a consolidated virtual server environment tends to generate highly-random IO that is very challenging for traditional storage architectures to handle–resulting in unpredictable storage latencies.
Exacerbating this risk, once the storage is shared, businesses lose visibility into the performance of the storage system for specific applications. This loss of storage visibility isn’t a big deal until application performance issues arise. At that point the lack of storage visibility makes root cause analysis more difficult and time consuming.
In addition, some application vendors still discourage clients from virtualizing the application the vendor provides. Their reasons may include concerns about possible performance issues and the cost of extra customer support hours that may be required to help a client troubleshoot any performance issues they encounter. In other cases their concern may center primarily on the loss of branded hardware sales.
Because of these risk factors, many firms have concluded that the risks of virtualizing critical applications outweigh the rewards. This in spite of the fact that business critical applications are precisely the applications where increased uptime, improved disaster recovery/business continuity capabilities, and faster application development would return the most value to the business.
VM-aware Storage Cuts Risk of Virtualizing Critical Apps
VM-aware storage cuts the risk of virtualizing critical applications by providing visibility into storage performance on a per-VM level. This visibility results in faster problem resolution and restoration of desired service levels. This visibility also enables a business to know how much storage capacity–including performance capacity–is available for additional workloads.
VM-awareness would make any storage system more manageable, speed up root cause analysis, and reduce the stress-level for storage administrators. But the greatest benefits of VM-awareness are achieved only when that awareness rides on top of a storage architecture that is designed for highly-random IO created by consolidated virtual machine workloads.
For example, Tegile’s Zebi Storage Arrays are designed from the ground up for the consolidated virtual server environment, and provide up to 200K low-latency IOPS to meet the demands of even the most critical business applications. Zebi VM-aware storage enables IT staff to both monitor and manage performance on a per-VM level. An administrator can (on the fly) “pin” the LUN associated with a critical application into flash memory; guaranteeing the lowest possible latency and the highest possible performance for that LUN.
The virtualized data center yields many business benefits, but concerns about potential storage performance problems have kept many businesses from virtualizing their most critical applications. New storage architectures that enable a business to monitor–and even manage–storage system performance on a per-VM level, such as the Tegile arrays, should enable businesses to confidently extend the benefits of virtualization to even their most critical workloads.
The Zebi HA2800 took top honors in the DCIG 2013 Flash Memory Storage Array Buyer’s Guide. The Buyer’s Guide may be downloaded for no charge with registration.