computing systems were built to be right-sized for a specific workflow and its particular requirements. Storage used to be relatively expensive, and even though prices have marched steadily and continually south, most of us still tend not to buy more than we need. With the media industry’s shift from proprietary storage hardware to commodity storage hardware, and with general-use PCs and servers becoming standard components in larger systems, these resources are becoming more commoditized.
Virtually every large media storage system today incorporates stand-alone compute technology from the likes of Intel and AMD. These components are often deployed as part of commercial off-the-shelf (COTS) systems from suppliers such as Dell and HP, because of the deployment cost savings over most build-it-yourself solutions. While this approach can significantly reduce the cost of processing power, it also tends to deliver much more power than is needed for the job at hand.
Often, even the most modest CPU available in a new storage system offers far more power than the file system database and access requests demand. Building to this standard is far more cost-effective for manufacturers than designing and building right-sized custom systems. In the realm of media storage and processing, the excess power in each server supporting storage system software translates to a waste of resources. One way to take advantage of this extra power is through virtualization, which leverages unused compute resources for applications that would otherwise require a dedicated server of their own.
From IT to the Media Production Environment
As an industry, media production doesn’t have the scale to drive change at the same level as more ubiquitous use cases, like IT server virtualization or database transaction processing, but that doesn’t stop the industry from borrowing the best of IT and leveraging it for its own purposes.
Through the use of virtual computing, media technology suppliers have created computational “safe zones” in which users can deploy applications and their requisite operating system environments. This model offers an extraordinarily less costly and less expensive alternative to spinning up an entirely new server, including its network switch ports, administration costs, and support plans, just to run any given application. Whether a small data mover, media browser, transcoder or render farm orchestrator, the application gains access to available resources on the overall storage platform. The media facility benefits from infrastructure that is more agile and more economical to operate, maintain and update.
Using virtualization and their knowledge of the compute power required by storage to run under any configuration or load, vendors can segment off processing power to support the storage system itself, and then divide and allocate the remaining processing power to support applications running on the same physical machine.
Thus, rather than set up an entirely separate server—and deal with associated hardware and software installation, updates, opera-tion and mainten-ance—to support even a small application, a
facility could instead deploy that application as a virtual machine running on an existing storage system.
Saving users a tremendous amount of time and money in taking applications live, this approach enables far greater flexibility in evaluating and rolling out new applications. In fact, it’s even possible to swap applications in and out at different points in the workflow, using the system as a media management server,
44