Distributing resources among applications often results in a situation where certain programs getting the short end of the performance stick. This leads to lag and other technical difficulties that are not permissible for mission-critical software, even more so in the cloud.
Analyst firm Wikibon says that while this phenomenon can be curved in a private data center, it is far more difficult to solve in a public cloud environment where a competitor’s app may run on the same piece of hardware as yours. The solution can only come from the host, as one cloud provider demonstrated in the latest Peer Incite meet-up.
Matt Wallance, the director of cloud architecture at co-location and IaaS service provider ViaWest, elaborates on how his firm managed to solve this problem by being clever about flash.
Wallance explained how his company approached multiple flash vendors only to get the same unsatisfactory response: the only way of silencing noisy neighbors is pumping up the IOPS and hope that there is enough to go around. It was SolidFire that eventually offered a viable resolution:
“SolidFire stood out because its tool set allows him to allocate a specific number of IOPS to specific customers or applications regardless of the demands of other customers. “With SolidFire the granularity is unique. Nothing else allows you to deliver SLAs in this way.”
Because IOPS can be allocated on the fly, SolidFire’s flash is often cheaper than traditional spinning disk when it comes to cooling, power consumption and even initial cost. But it’s the performance that really sealed the deal: Wallance says that 2-4 milliseconds is the worst ViaVest has seen for the most demanding instances.
For the full use case and analysis, check out the full one-hour session below.