Best Effort-as-a-Service Just Won’t Cut It
During the recent Hosting & Cloud Transformation Summit in Las Vegas we hosted our third QoS Summit roundtable, this time with Tim Stammers from 451 Research. We were joined by representatives from both service providers and large enterprise IT departments and debated the opportunity and challenges of delivering storage Quality of Service to their end users. The most interesting discussions of the evening was about the complexity of turning raw IOPS into a service that could easily be consumed by end customers and internal business units.
When it comes to discussing the merits of a storage system, IOPS get the most airplay, but IOPS alone don’t directly translate to benefits for end customers. It is not just raw performance, but rather the ability to deliver guaranteed performance which can be clearly delivered to end users. Without the ability to guarantee a certain level of performance all you are providing “best effort”. The line of the evening came from one of the attendees who commented on what you end up with without the ability to guarantee performance. He stated, “without quality of service all you are is a ‘best effort provider.’”
For any CxO looking to host production sensitive applications in a shared infrastructure, either in the cloud or on premise, best effort just simply won’t cut it. But delivering storage QoS is much easier said than done. In addition to the limitations of any legacy infrastructure to deliver guaranteed performance, service providers are challenged to properly package the concept of performance to end users.
End users care about applications running smoothly and how many IOPS it takes to make that happen is a foreign concept to most. As a result, exposing a bunch of IOPS options to end users is often not the answer because it causes confusion and unnecessary complexity. The more sensible starting point echoed by our QoS Summit attendees is to expose simple tiered services with different levels of IOPS bundled in. Over time these services can evolve to more granular offerings as customers gain a greater understanding of different performance levels/tiers needed for different applications.
Unfortunately, the ability to present a spectrum of services from a single storage infrastructure has not been possible previously. So what often results are basic block storage services delivered from one platform and advanced higher performance offerings from another. From an admin perspective, an application starts on a lower performance tier and movement to a higher performance offering requires all sorts of manual administration and data migration. Along the way none of this performance is actually guaranteed, it is best-effort. This hassle removes any sort of agility on the part of the service provider to confidently respond to the evolving needs of their customers. This is where the adaptability of a storage system is critical. This is where SolidFire comes in.
At SolidFire we have designed our architecture such that storage performance and capacity are composable resources than an administrator can scale up or down based on the needs of the application. With the ability to allocate and dynamically adjust IOPS and GBs at a very granular level, administrators can provision basic storage tiers to start and simply adjust them over time as the needs of the applications and end users become more apparent. Using SolidFire this adjustment is just a simple API call, no downtime, no performance unpredictability, no data migration required. We believe this kind of flexibility is imperative for service providers and enterprises looking to transform IOPS into a recognizable benefit for their end users. Those around the table at our QoS Summit seemed to agree!
For more detail on our presence at this year’s HCTS event you can watch our CEO, Dave Wright during his panel discussion, Profiting from Cloud Storage in an Era of Software-Defined Everything. The slides from the panel can be found here, while the video from the discussion is available below.
-Dave Cahill, Director of Strategic Alliances