2014年11月25日
By 蒂姆Siglin 特约编辑
专题文章

Elemental Pushes Concept of Software-Defined Video

基本技术, known for higher-end encoding and transcoding appliances that rely heavily on the parallel-processing capabilities of a graphics pocessing unit (GPU), is on a mission: Use software to define overall video workflows, 从收购到交付.

Sam Blackman, co-founder of Elemental, recently spoke with StreamingMedia.com about the ideas around the intersection of networking and video. Blackman has has been pushing the concept of software-defined video (SDV), which sounds similar to—but is quite different from—the more popular software-defined networking (SDN) term.

在SDN的世界里, there’s a commonality—or six of them if you adhere to the Open Networking Users Group (ONUG) model that we will cover in the January 2015 issue of 流媒体 magazine, as part of an article on next-generation enterprise video platforms—and ONUG’s "构建共同网络生态系统的六个步骤" outline what product manufacturers need to provide in order to fit into the larger SDN ecosystem.

In the world of streaming media, the software-defined video (SDV) term is a bit less defined.

我问布莱克曼什么是共性, 从元素公司的角度来看, a vendor would need to provide to fit into an SDV ecosystem.

Blackman started out with a definition of SDV. “Software-defined video is software code running on off-the-shelf hardware (CPU or GPU) that performs a video function that previously required specialized hardware (ASIC, FPGA, 或DSP),他说.

The distinction of using general-purpose computing (GPP) components such as the CPU or GPU, rather than the use of more specialized digital signal processors (DSPs) or other purpose-built chipsets is not unique to streaming media. But the approach Elemental espouses is somewhat unorthodox.

在我们的案例中, 我们正在进行压缩, 包装和交付完全在软件中,布莱克曼说. “SDN领域的类比是芒网络, which has built a Linux-based stack on top of off-the-shelf hardware to provide an extremely high-performance, 灵活的软件组网交换机.”

Blackman notes that SDN is several years ahead of SDV. “We see some of the larger video infrastructure vendors starting to build interfaces that allow them to natively integrate SDV products into their broader portfolio,他说.

He then listed two example architectures, the Ericsson's 夏娃 (爱立信虚拟化编码)和思科的 V2P (Videoscape Virtualized Video Processing) architectures.

“我猜测，随着生态系统的成熟,布莱克曼说。, “rather than specific vendors creating proprietary interfaces for SDV compatibility there will be mutually-agreed upon open standards that insure interoperability.”

This agreed-upon set of open standards has not, 正如布莱克曼提到的, been agreed upon as it has in the networking world, where "big iron" companies like Cisco and 瞻博网络 both have their own definitions of how to interoperate, but which are also held to a standard baseline of interoperability if they choose to make their products ONUG compliant.

An example of how far we still need to go to get to an SDV world, 虽然, 来自一家大型钢铁公司瞻博网络他的朱诺斯产品赢得了一项大奖流媒体读者选择奖 has created one standard for transparent caching, while several other competing companies in that space have provided their own proprietary transparent caching technologies.

In terms of how SDV and SDN fit together, Blackman says that’s a fairly straightforward scenario.

“SDDCs will entirely consist of virtualized infrastructure,布莱克曼说。, referring to virtual machine solutions in the emerging software-defined data centers field, “包括计算, 存储, 和网络.”

“SDV applications will ride seamlessly on top of these VM resources,” said Blackman. “Media companies managing these SDDCs will be able to spin up and down SDV resources (VOD transcoding, 线性或实时编码/转码, 包装, and content origination) as business needs dictate.”

最后, the intent is straightforward: enable more flexible, 可伸缩的, and economically optimal workflow models than we’ve had to date. Blackman says that is better for any media customer.

“It will enable our customers to be much more nimble in a time of unprecedented change,他说.