Tiered Storage Benefits Achieved Non-Disruptively By Doug Rainbolt On February 22, 2011
The promoted benefits of automated tiered storage include providing an automatic means to place data, based on its importance, on the right media. The three most common examples of “right” media promoted by some of the mega storage vendors include SSD Flash for one’s really hot data, Fibre Channel drives for hot data, and SATA drives for not so hot data. Underlying this automation is both sophisticated software technology that analyzes data and/or enforces policies for data placement and available CPU cores to execute it. The designers of such systems will say that they literally spent hours analyzing real customer data, getting a sense for access patterns and performance. I don’t know about you, but this strikes me as being very complex. How do we know that they have it right or that the data they’ve profiled matches my world, which may include diverse data and I/O patterns?
It would be interesting to analyze one of these automated tiered storage arrays up close. When I say close, I mean dig down into the storage controllers and understand how the processors, memory, and I/O are working. Where are the system bottlenecks? Every system has them. The more “technology” you throw at a storage controller, the more contention there will be for resources and there is a very good chance that the controller will become the bottleneck long before the disk drives or Flash. And when the controllers are saturated, it doesn’t matter how much media you have or what type, you’re going to see performance degradation. How this ends up manifesting itself for a customer is limited scalability. The vendor’s answer? Throw more nodes into the cluster or add another scale-up system. But these are often band-aids that mask the real problem.
At Alacritech, we think that it’s possible to achieve the benefits of tiered storage without actually having tiered storage and the associated complexities and costs. Users can non-disruptively install an ANX 1500 NFS Throughput Acceleration Appliance alongside their existing NAS infrastructure and benefit from active hot data being served to clients while using less expensive media on the NAS. The ANX’s sole purpose is to cache hot data in a very scalable way, benefiting many NAS devices. Our simple philosophy is to deliver a performance layer to complement the NAS capacity layer. These layers are focused on different tasks, each optimized for efficiency. Together, through your real-life data access patterns, data placement is optimized automatically, costs are kept low, and life is simpler for storage administrators.
Related posts:
Leave a Reply