Intel recently published a case study on a recent cloud deployment at 3M that is worth a read as it is a particularly good example of how Panasas® ActiveStor™ can be profitably used for manufacturing workloads in a private cloud context. SGI is a valued Panasas partner and is in a strong position when it comes to offering a total solution to overcome big data challenges. In 3M’s case, they deployed a fully integrated, policy-driven solution comprised of 52TB of Panasas ActiveStor primary storage and 150TB of SGI ArcFiniti™ archival storage. Networked storage is an industry-wide best practice for private clouds, with the flexibility to make file systems available on any or all virtual machines operating on the servers, in this case SGI Rackable™ and SGI Altix® UV models.
Why Panasas storage for manufacturing workloads? In short, performance and ease of management. Every ActiveStor shelf of storage delivers up to 1500MB/s of throughput, scaling quickly to tens or even hundreds of gigabytes per second from a single file system as additional shelves are added. This means that the performance of manufacturing design applications like ANSYS, Abaqus, Accelrys, and Fluent used at 3M doesn’t have to suffer from a lack of storage I/O as is often experienced when using legacy NAS systems. Finally, commercial HPC users need a big data solution that is easy to scale and easy to manage; both strengths of the Panasas solution.
3M Systems Architect Peter Bye put it this way: “In the past, we got requests from folks who needed to run something that we couldn’t do without making some serious architectural changes or purchasing additional hardware. More often than not, we wound up having to buy hardware and custom-configure it for a project that might only last six weeks. With the new system, we can give them a lot of flexibility for experimentation, and the resource load on our team for those requests is reduced to almost zero.” The end result is that 3M’s HPC users can now refine designs more quickly, leading to improved product quality and engineering productivity.