Newswatch June 2012

 

SMPTE Industry News - Monthly Tech Focused Newsletter of the Society of Motion Picture and Television Engineers

 

Hot Button Discussion
High-Speed Networking
By Michael Goldman

As media companies dive deeper into virtual production, collaboration, and distribution, networking methodologies are pushing further into the media world, while formats, codecs, and technological innovation continue to evolve at a startling rate. Today, there are numerous ways to move video and audio files for production and distribution, which poses challenges for media companies as they sort through their networking options.

Security is obviously a chief concern, along with speed, efficiency, and automation. But content creators and users are continually demanding more innovative capabilities from wide-area networks as their businesses move into a more dynamic collaborative environment.

According to industry veteran Don Craig, former co-founder and CTO at Omneon Inc., now an independent consultant and SMPTE Fellow, exciting developments are well under way for wide-area networks capable of delivering more media, services, and applications in a cost-effective and automated manner.

 

"A high-performance network can mean many things—to me, it means a specialized network capable of carrying high bandwidth essence and metadata, possibly including a video clock, all the way through a production chain in a way that is transparent to the people receiving the media or pieces of media," Craig explains. "The network supports creating content and then putting it through a variety of authoring and mastering processes all the way to final distribution. Content used to be carried as bits of essence on a point-to-point wire, but now, it is contained in a file or stream carried by a protocol over a general-purpose, point-to-multipoint network. That network has to be provisioned to meet worst-case bandwidth and latency expectations of the demanding media user.

"In the meantime, IT network topologies become more dynamic and complex with the introduction of virtualization. Link and switch speeds are higher, and overall throughput expectations increase dramatically. But the problem of managing and configuring routers to support a desired virtual topology becomes extraordinarily difficult. The distributed routing algorithms used in IP networks will often produce a route that is not quite the one desired, and it takes careful tuning to achieve acceptable performance. The introduction of virtualization makes routes dynamic, which in turn, disrupts any tuning that has been done. There's a new notion known as a Software Defined Network (SDN), which relies on separation of the so-called data planes and control planes in an IP router. Rather than establish packet routes using distributed routing algorithms, routers query a central resource to establish what's known as a 'flow.' This central resource can make decisions far more rapidly than a distributed routing algorithm, and can employ arbitrary criteria for establishing a route."

One of the major developments in this arena is the research and promotion of the SDN solution by the Open Networking Foundation (ONF)—a non-profit consortium focused on "the transformation of networking through the development and standardization of a unique architecture (SDN), which brings direct software programmability to networks worldwide," according to the ONF's mission statement.

 

ONF work has produced what the consortium refers to as a "switching ecosystem" supported by the OpenFlow interface. OpenFlow, according to Craig, essentially allows a centralized computer brain to configure all the data flow behaviors within a network. OpenFlow is implemented by an open source code base that can be used and experimented with as it develops, with the eventual goal of developing it into an industry standard. OpenFlow separates the fast packet forwarding data path from the high-level routing control path in the network. The data path continues to reside on a traditional switch, while monitoring and adjusting of routing decisions resides with the OpenFlow control server. 

"The last few years of networking have seen a dramatic increase in the use of virtualization, especially in big data centers or clouds," Craig says. "This means you are not talking to a real computer, but rather, an emulated one, which can move from one physical computer to another. It's the network's job to find that emulated computer, but IP network architectures are not designed to find hosts that move frequently. So there was a need for an automated system to configure routers to support virtualization. OpenFlow, originally developed as a testbed for research protocols at Stanford, was adapted to meet that need. Now branded as SDNs, such networks exist alongside the existing IP network capability of routers. It's an important development."

Craig adds that the expectation of this work is that "SDN system architecture can make it easier to create networks that guarantee provisioning for video and audio transport—in effect re-inventing the video crossbar switch, but in a virtualized, packetized environment." (Craig recommends reading an ONF April 2012 White Paper on the subject, which you can find here.) 

Part of the evolving network paradigm, of course, involves cloud computing. Craig suggests that a wide range of workflows can now incorporate the cloud. It can be used, for instance, to store or archive content, to communicate content, or to find content, but then deliver it with some form of peer-to-peer transfer. "That technique eliminates cloud security concerns by keeping the actual content within a managed point-to-point link," he suggests.

 


Join Tektronix in our free video seminars to learn about the key measurements and test techniques for Digital TV, MPEG Test and File Based Workflows. Simply click here to register.

Despite the widespread adoption of IP networking technology for video transport, however, SDI over coaxial cable is still broadly used in professional infrastructures. "The biggest changes to SDI lie in the augmentation of metadata support and the addition of a broad range of new image formats," Craig adds. He also points to work being done to move 3 Gbit SDI carrying various 3D formats or 4k formats over different kinds of networks. (For more on SDI enhancements, he recommends reading the2011 SMPTE Conference paper by Nigel Seth-Smith of Gennum.)

But for wide-area networking requiring greater bandwidth over longer distances, in particular, Ethernet has evolved beyond enterprise network applications and into the world of carrier/service provider/WAN networks—typically carried over Optical Transport networks using multiple wavelengths per fiber, Craig suggests. He also points to another 2011 SMPTE Conference paper by solutions architected Michael Watford of the Ciena Corporation. In that paper, Watford describes a research project that deployed a wide-area transmission system for 3 Gbit SDI feeds over 10-and-higher gigabit carrier-class Ethernet to link hockey arenas in Switzerland in a dedicated video network.

Watford's paper also discusses a process of mapping SDI signals into Ethernet frames, similar to how the SMPTE 2022-6 standard maps SDI into IP packets using the Real-Time Transport Protocol (RTP). "There's a trend to carrying all those SDI standardized formats over IP or just plain Ethernet," Craig elaborates. "Carrier class Ethernet works well for wide-area carriage of television signals."

Indeed, Craig foresees a migration toward Ethernet for certain types of networks.

"More and more television signal processing and interconnection is becoming generic in terms of the use of IT hardware, rather than special-purpose television equipment, and SDI comes from the special-purpose television equipment part of the business," Craig explains. "And so, as time goes by, we're going to be migrating from SDI to Ethernet for more and more connections."

"Where SDI is still predominant is in professional applications over relatively short ranges where coaxial cables are available. There, it's the inter-connection choice, because it's simple. But as more capabilities are added to Ethernet for managing latency, as well as bandwidth, and as Software Defined Networking makes it easier to set up a managed communications path, I think Ethernet will continue to make inroads.

"We are also seeing new consumer interconnect technologies used in professional applications. HDMI, used primarily for piping video to modern televisions, is now found in inexpensive production applications connecting cameras to production switchers with low-cost optical repeaters. And conversely, know-how originally acquired for professional SDI applications has found its way to the consumer space (in the form of) the new Thunderbolt interconnect deployed by Apple, which uses chips inside its active cable connectors developed by Gennum."

Craig adds that Ethernet and IP packet formats are very stable, particularly with IPV6 (Internet Protocol, Version) deployment proceeding rapidly. 

"And Ethernet packet headers have not changed much in the last decades," he says. "As you go up the network layer stack, toward the applications, there is a jungle of codec and wrapper formats. However, the need for interoperability and reduced complexity is driving us to converge on some limited number of formats. As our understanding of the technical needs of our entire community improves, particularly in the highly complex area of production workflows, we are able to replace multiple old formats with fewer, more capable new ones."

And speaking of production workflows, Craig calls the production industry's network demands "quite bursty"—almost every piece of work is deadline critical, and network loads rise and fall depending on those deadlines and production cycles. 

"As well as provisioning their own networks, studios buy bandwidth and connectivity from outside suppliers to meet their peak needs," he explains. "They have to manage those networks with software that shows the demand being placed, and assists in prioritizing and queuing the traffic. At that point, it's about economics. The technology to build fast and reliable dedicated networks exists, but low latency and high bandwidth cost extra, even when you are not using them. Coupled with the exacting reliability and security demands of media companies, the cost-effective design of networks for capture and transmission of media content remains a challenging problem."

Comment
back to top

 

 

News Briefs
U.S. Regains Supercomputing Top Spot

Thanks to IBM's Sequoia Tower, technology currently residing at the Department of Energy's Lawrence Livermore National Laboratory, the U.S. has regained the lead as creator of the world's fastest supercomputer. According to a recent TechNewsWorld report, Sequoia recently claimed the top spot on the top 500 list of the world's fastest supercomputers released at the 2012 Supercomputing Conference in Germany. Sequoia overtook Japan's Fujitsu K Computer (located at the RIKEN Advanced Institute for Computational Science in Kobe, Japan)—marking the first time since late 2009 that an American supercomputer held the top spot. Sequoia achieved 16.32 petaflops on the Linpack benchmark measuring system (a method of measuring execution speed through a series of complex equational tests) using 1,572,864 processing cores to earn the designation, while the Fujitsu K system hit 10.51 petaflops using just over 705,000 processing cores. The rankings are done twice a year, and before the Japanese system, two Chinese supercomputers had achieved the top spot over the last two years. While the rankings, which have been published for 39 years, may seem more about bragging rights, the article points out that they are an important tool in bringing business and research dollars to various scientific institutions around the world.

Comment
back to top

 

The Wonders of Nanoplasmonics

The nascent, little understood, under-reported, yet amazing science of nano-technology marches onward. The latest development comes out of a report from researchers at the University of California, San Diego, Jacobs School of Engineering. The report states that a method for allowing metallic nanocrystals to self-assemble into larger, complex materials has been developed, that, potentially, could be used to manufacture antennas and lenses, among other things, in the future. The research emerged from the field of nanoplasmonics, which is about creating new materials to manipulate light through structures that are smaller than the wavelengths of light itself. The nanocubes used in the study, for instance, have the thickness of 0.1 microns—far less than the width of a strand  of human hair. If oriented correctly, the cubes can reportedly confine light (for an antenna configuration) or focus light (for a lensing configuration) at different wavelengths. You can check out the findings in the Nature Nanotechnology journal here.

Comment
back to top

Too Slippery to get Wet

An interesting example of the synergy between scientific research in other disciplines and how they can be applied in the broadcast world is offered in a recent report from the TV Technology site, which postulates the idea that recent research at Harvard University could have applications for helping broadcasters find a low-cost/low-power way of de-icing transmitting antennas. Scientists at Harvard, according to the report, have been developing a new surfacing technology designed to be so slippery it literally repels anything, including water and ice. The technology is called Slippery Liquid Infused Porous Surfaces (SLIPS) and could lead to the creation of a liquid-interface coating for metallic and other surfaces so that they simply can't ice up under any conditions. To learn more about the Harvard research, check out this report from Harvard's Wyss Institute for Biologically Inspired Engineering. Researchers say the idea came from a biological source—the water-repellent lotus leaf.    

Opinions expressed in SMPTE Newswatch do not necessarily reflect those of SMPTE. Reference to specific products do not represent an endorsement, recommendation, or promotion.

Comment
back to top

 

Let us know what you think about SMPTE Newswatch! Provide suggestions and feedback on improving this newsletter by completing this brief survey.

back to top

Table of Contents

High-Speed Networking

U.S. Regains Supercomputing Top Spot

The Wonders of Nanoplasmonics

Too Slippery to get Wet

Connect with Us

 

SMPTE Events

22-25 October 2012
2012 SMPTE Annual Technical Conference & Exhibition
Hollywood, CA

Regional Seminars

Thursday 27 September
Montreal, Canada

Thursday 8 November
New York, NY

Saturday 10 November
Toronto, Canada

Thursday 15 November
Orlando, FL

 

Get more information about SMPTE.

 

 

To advertise in this e-newsletter, contact Jeff Victor at jeffvictor@comcast.net or 224-436-8044. 
Eletter created by Cavich Creative, LLC