O ne of the things I like most about the technology industry is that it is characterized by a relentless quest for improvement. There is always something newer, faster and better right around the corner. Never a dull moment. And always the drive to be first and best.
Last week, for example, the first commercial non-stop flight from Australia to the United Kingdom took place. A 787-9 Dreamliner departed Perth and landed at London Heathrow 17 hours and 9,009 miles later. Thanks to more efficient engines and aircraft design, this elimination of a stopover resulted in a nearly 20% time savings from what was already a very quick journey (historically speaking). That said, aircraft manufacturers are busy engineering machines that will enable even faster journeys, and soon there will be non-stop service between any two points around the globe. The question, of course, is not if that will happen, but simply when.
In our industry, the story couldn’t be more similar. Just a few years ago, the introduction of SSD’s as a storage medium introduced a sharp increase in storage performance that yielded significant benefit to many members of our community. Suddenly, editing and rendering 2k 30fps media seemed less daunting. But these first SSD’s leveraged legacy SATA/SAS physical interfaces and protocols, which meant the CPU couldn’t make the most of this faster, yet more expensive, storage device. That might have been fine if we were frozen in time from a production standpoint, but, of course, the industry wasn’t going to stop at 2k, as 4k was just around the corner.
With NVMe, much of this inefficiency, and therefore unrealized performance, has now been addressed by introducing a highly scalable, high-performance storage protocol that has exceptional low latency paths to the underlying storage media. Thanks to this new optimized interface, applications are rarely starved for data, and editors can confidently work on the super high resolution / high frame rate media that is all too common today.
The one catch in making the above possible is that this often required the storage to be local to the client, which ultimately required each workstation to be kitted out with a high number of SSDs, introducing significant cost overheads. And if the workstations are working on content independently, collaboration also becomes an issue. So, although NVMe is a great solution for certain media workflows, we at Quantum were looking for something to make the technology even more useful and cost-effective for a greater ecosystem of customers. Thankfully, there is a new kid on the block (technically speaking) that we are leveraging to make that happen.
As one of the many initiatives we have around making the most of flash-based or flash-augmented hardware, Quantum recently formed a partnership with Excelero to introduce a new storage solution based on NVMe over Fabrics. The magic of NVMeoF is now multiple workstations can utilize pooled NVMe storage devices across a network at local speeds and latencies, thereby providing exceptional streaming and IOPs performance. The underlying storage is presented as a dynamic block volume, yet the entire environment is tightly integrated with Quantum’s StorNext file system, which provides the benefit of having all data accessible by multiple clients in a global namespace. In a way its the best of both worlds — extremely high-performance storage, yet with the cost-efficiency of sharing resources, and finally the ability to access data in a coordinated way.
And to highlight this capability, this year at NAB we will be featuring an exciting demo with Quantum Xcellis® leveraging the capability of Excelero NVMesh to drive performance of a workstation editing 8K streams, all using our StorNext file system to intelligently manage content from ingest to delete.
I’m sure you will be busy at the event, but please take a moment to stop by and take a look. We promise it will be well worth your time!