We all know the Media & Entertainment industry has specific needs that involve working with small user groups, large file sizes and sequential file access. Because of these unique requirements, traditional IT hardware and software solutions struggle to sustain the required performance needed for post production solutions.
This is a two-part problem:
The first challenge stems from a lack of file system intelligence. Every computer loses performance over time; defragmentation routines performed on a regular basis will help to restore performance for awhile, but the problem will occur again after a couple of months…or even weeks. Why does this happen?
Standard file systems do not have the built-in intelligence needed to handle defragmentation—especially when handling media files. In order to obtain the best possible performance from your system, the files should be in a “perfect layout.”
Perfect file layout entails writing one big chunk or one piece of data, located where the system can gather up the data in the least amount of time.
Unfortunately, that‘s not what standard file systems do. Instead, they break up the files into chunks in order to make maximum use of the available free space; this results in files being fragmented.
But that’s not all; as soon as multiple workstations write to the same file system at the same time, files unavoidably get interleaved on the disks, as the file system saves the files in the order they were written. So sequential files become separated. And that’s still not all!
As workstations send multi-threaded writes to the file system, the already-fragmented and interleaved files now get randomized. This is because multiple threads send files out of order, hence they arrive randomly on the disk.
The second challenge is due to limitations in storage performance. How hardware is designed can vary from vendor to vendor. And when a vendor lists performance specifications, too often they are giving peak specs that are generated by testing equipment configurations that you don’t need or cannot afford.Do not confuse traditional performance metrics like throughput or IOPS with what is really required for M&E workflows: streaming performance.
Fast throughput or high IOPS will not provide you with the performance you need in a workflow with real-time access requirements. In a typical post production workflow, there will be multiple clients attached to storage, and at times those clients will all be accessing that storage simultaneously.
When concurrent data access occurs, traditional metrics of MB/s and IOPS will not answer the question: “Do I have a enough performance for this set-up?” Determining the answer to that query requires knowing several key factors: the number of streams, their resolutions and frame rates, and if multiple streams are writing simultaneously to the storage.
Finding a solution to both the file system and hardware issues mentioned above will lead you to more money and make you a happy customer:
Recommended Post-Production Storage Solution
For more information on how to find a solution that meets these needs, learn more about the Bright Technologies solution with Nexsan.