The PCIe (PCI Express) standard will have a disruptive impact on the business and workflow processes in video/production/post production/etc. Leading storage vendors are catching on to the idea as we speak. The industry currently faces technology challenges in dealing with the continual purchasing of storage equipment to keep up with the rising needs of performance and archiving – and how the PCIe standard will be the enabling tool to keep capital and operational costs down.
The PCIe standard, prevalent in all aspects of equipment for post – is already changing the performance/price landscape. Implemented into the storage infrastructure – the PCIe standard will dramatically improve performance and workflow efficiencies, as well as lowering the total cost of ownership due to the dramatic decrease in footprint, heat, cooling needs, etc. This coupled with massive price decreases will put InfiniBand, Fibre Channel, SCSI, etc. standards based equipment on notice. It will also allow more 'closet freelancers' enter into the mid-high-end post because of reduced equipment costs. Simplicity is also a factor. A user on any level needs only plug in a PCIe storage solution into a PCIe slot - and instantly, they'll be able to handle multiple streams of 4K, 2K, HD, etc. workflows in real time.
Background on PCIe (technical):
PCI direct attached storage has been possible for many years, but was previously cost prohibitive for lack of standard hardware and the complexity of the parallel PCI bus architecture. With the release and common implementation of the PCI Express (PCIe) standard, the natural migration to PCI attached mass storage device arrays has become easy and very cost effective - especially with companies like JMR Electronics (www.jmr.com) leading the way in development with their BlueStor(TM) PeSAN(TM) technology.
The reason to use PCIe attached storage arrays is very simple: This approach brings the storage devices closer to the computer CPU by eliminating bus bridges, along with the overhead and bottlenecks they create.
Using a dedicated Xscale processor for each PCIe attached storage subsystem and connecting to the host via multilane PCI Express frees CPU resources in the host (server) to improve its performance and ultimately the overall data transfer rates of the entire system.
Since PCIe is switched architecture it can be scaled indefinitely and adding more external Xscale processors (in the form of low cost hardware RAID controllers collocated with the disk drive subsystems) and more disk drives continuously improves performance until the host PCIe bus is saturated. Currently, that occurs at 4000 MB/s using first-generation eight-lane (x8) PCIe; however, second generation ("Gen. 2") PCIe is emerging right now, along with common availability of x16 PCIe mother boards: This allows bus speeds of up to 16,000 MB/s (16 GB/s). Thus, at a per-disk optimum data transfer rate of 100 MB/s, one can inexpensively scale storage to 160 disk drives (at 1TB each = 160 TB) without overburdening any host computer resources. As PCIe standards continue to increase bus speeds and bandwidth, and disk drive capacities continue their endless improved density path, it is easy to see that in 2009, using Gen. 2 x16 PCIe architecture and 1.5 TB SATA disk drives, one can build a 240 TB (1/4 Petabyte) array capable of >10 GB/s transfers – and do so, inexpensively! Current cost models validate mass storage system costs well below one dollar per GB using this new approach to attached storage.
For more technical/lab test information, contact Steve Katz: 818.739.1112 stevek@jmr.com
