Demand-Driven Storage Blog

Today we announced the FXT 2700, which is our first product based on SLC Flash technology. In the time that customers have been evaluating this product, we have talked to many organizations interested in testing the benefits of SSD technology. The good news is that for those looking to experiment with the performance benefits of SSD, the FXT 2700 turns out to be an easy and cost effective way to add SSD performance to any NAS-based application.

Using Avere’s 2-stage implementation, the FXT 2700 can be deployed in any vendor’s environment. Deployment and removal requires no down-time. You are not limited to the offerings of a single vendor. You are not limited to expensive internal hardware adapters, nor are you limited to entire arrays of SSD media that must be allocated to specific volumes.

The FXT 2700 only allocates the portion of the namespace to the SSD media that needs the added performance. The SSD media is shared for the entire namespace. It can be deployed in any environment with an existing NFSv3-based NAS server. And the SSD media can be scaled by clustering FXT 2700s which globally share the media across all nodes and all application workloads.

Avere built the FXT 2700 specifically to support two types of workloads, those with very high performance applications and those applications with high levels of random-access data.

For high performance applications, our typical customer is already running a NAS server, usually with costly Fibre Channel media by necessity. Introducing an FXT 2700 cluster in that environment dramatically reduces the workload on the NAS server. In some of these deployments, short stroking is used to eke out even more performance. By reducing the workload on the NAS server, the requirement for short stroking is eliminated, which effectively frees up previously unusable storage space on already spinning media.

For applications with high levels of random-access data, hard drive based NAS servers can be inefficient. Hard drives are excellent at reading data from their platters once the read head is located at or near the proper track. If the read head is not near the proper track, which is usually the case for random workloads, the head seek time reduces the performance of a hard drive from thousands to hundreds of operations per second. Adding the solid state tier of the FXT 2700 in such systems not only improves application performance, but it also allows the hard drives to be used for bulk updates, which increases their efficiency as well.

Through the FXT 2700 launch and customer evaluations, I continue to be fascinated by the many applications that users have chosen to deploy its SSD technology, even outside of our expected workloads. This breadth of applications and workloads that can be attributed to the FXT 2700′s ease of deployment, made possible by Avere’s vendor neutral 2-stage architecture, and by its cost effectiveness, in that the SSD tier is automatically used only where needed.