First there was server sprawl, then there was storage sprawl and now it’s “performance sprawl”. Performance sprawl occurs when there are multiple pools of flash storage spread out across the data center. For example, many organizations have flash resources inside their server infrastructure, as well as flash devices inside their shared storage platforms. In some cases, network based flash is deployed in the storage infrastructure itself. The problem isn’t necessarily that there are disparate silos of flash, but rather there is no effective means of centrally managing all of these resources.
Through technologies like server virtualization, many businesses are enjoying the benefits of consolidated infrastructure. But performance sprawl threatens to upset the “software-defined” apple cart by creating additional points of flash hardware management throughout the storage infrastructure. In fact, as we discussed in our webinar “Stopping Performance Sprawl”, some organizations have even resorted to hiring a dedicated storage performance specialist to help remedy this situation. While having additional storage expertise in-house isn’t necessarily a bad thing, in this scenario, it’s kind of like the flash tail wagging the performance dog.
Single-Point of Flash Management
To get out from under the performance sprawl management headache, organizations should consider flash acceleration software technologies that can centrally manage flash wherever it resides in the data center – whether it’s server-side, array or network based flash. This can help to reduce management overhead and actually improve utilization of all flash assets in the data center.
For example, in addition to providing a single-point of performance management, some solutions enable disparate flash resources to be stratified across the data center for “multi-level” or tiered caching; whereby the lowest latency, highest performance resources can be reserved for the most active application data sets.
Other practical ways to centrally manage and leverage multiple flash resources is through “remote caching”. This feature enables the seamless migration of VMs (server vMotion) without causing noticeable degradation to virtualized application performance.
To learn about all of the benefits of implementing a single point of performance management for data center flash resources, please watch our on demand webinar with Intel. By watching the webinar, you can download a copy of the Storage Switzerland article,“How To Choose The Right Caching Architecture”. Once you register on the Storage Switzerland BrightTalk channel, you will never have to re-register. Plus you will have immediate access to our library of more than 60 on-demand webcasts.
Intel is a client of Storage Switzerland