Data Center Tech Blog
Victoria Koepnick

Product Marketing, Software Solutions

The key obstacle to virtualizing mission critical applications has always been performance. By definition, server consolidation means running multiple workloads on a single shared hardware platform. While todays CPUs and memory provide the raw compute power to handle most virtualization workloads many storage systems are now the performance bottleneck. The reason why storage systems are now the performance bottleneck is due to I/O latency.

There are several options for addressing the storage performance bottleneck – from adding more spindles to existing arrays, deploying hybrid arrays which include spinning disks and flash drives, to all-flash arrays.  But the storage system can only go so far to address the bottleneck because they do not sufficiently eliminate the I/O latency. These options can also be expensive and disruptive to existing IT data management and data protection policies.

Perhaps the most effective way to address the performance bottleneck is not in the storage, but in the server. Deploying flash directly in the server can eliminate much of the I/O latency and coupled with software that can automatically identify the data that would most benefit from being placed on the flash offers a seamless solution.  There are several benefits to deploying this type of solution:

Deploy the amount of flash precisely where it’s needed. Flash in the storage architecture is a tide that raises all boats. Flash in specific servers delivers the benefits to those servers precisely, and even the software can target the storage volumes and virtual disks that are most important to your business needs.

Choose the flash you need. An advantage of a software-based approach to managing server flash is that you can select the flash that suits your needs, in terms of performance, capacity, server compatibility and budget.

Move storage I/O traffic off storage and into servers. When you deploy flash storage in servers, you don’t just improve application performance. You move the majority of your storage I/O traffic off your storage network. That means that you don’t just get faster “storage performance.” You get greater capacity and longevity from the systems you already own.

This is the key to the hidden value of server-based flash storage: You can postpone very expensive expansions, upgrades and overhauls of existing storage systems through a far smaller expenditure on server flash and software. And – let’s not forget – you’re doing this while delivering the best solution for your organization’s application performance needs.

Leverage existing storage investments and strategy. The purpose of server-side flash should be to complement existing storage systems, so you can add flash performance to your storage without replacing your existing storage. On the one hand, the cost of acquisition alone is reason enough, but an additional benefit for many organizations is that this non-intrusive solution means they preserve existing IT processes and policies that an architectural change would have disrupted.

Good news: you don’t need to endure the pain of a forklift upgrade to your storage infrastructure to eliminate I/O latency and virtualize mission critical applications.

For Ethan Erchinger, VP Operations and Solution Architecture at Plaxo, a Comcast Silicon Valley Innovation Center and SanDisk® customer, he wanted the efficiency and savings of virtualizing his mission critical database applications, but couldn’t risk performance degradation. By deploying a PCIe flash device and FlashSoft software in his servers he was able to virtualize mission critical applications, consolidate server deployments, and avoid a costly storage upgrade.

Ethan estimates that through this initial deployment of FlashSoft he will realize a savings of nearly half a million dollars.

 

 

 

 

 

 

Read Plaxo, a Comcast Silicon Valley Innovation Center, Discovers the Value of FlashSoft for more details.

SSDs in the Data Center – The Big Flash Questions

Meeting the Big Data Challenges: Scale, Performance, Reliability and Costs

subscribe blog