To hear Cisco tell it at its Cisco Live conference, the next-generation network will be nothing but video. But if you are hoping to offload video as a cloud service, think again, says Guido Jouret, Cisco's vice president of enterprise video and CTO of emerging technology.
"Video and cloud in some cases do go together, but in some case are polar opposites," he says. "Even if you as an enterprise had no concerns about taking your internal sensitive videos and posting them to some cloud video service like YouTube, it would still be a bad idea," he says. And that's because all the people watching those videos will be your employees. Every time they press play on the video, that video will have to stream through your Internet gateway and over your skinny corporate WAN link. "Talk about single point of failure and chokepoint. So, topologically it's a really poor choice of where to put the video. With video, because it's so big and so sensitive, topology matters. That's not the same with hosted email. Email is skinny."
Dare we point out that this point of view assumes that your enterprise is organized in a classic all-employees-are-located-behind-the-firewall-at-headquarters scenario. The more distributed your workers, and your network, the less of a chokepoint a single Internet gateway would be for hosted video.
In any case, the overall concept of where you place the video is an important component to how you prepare your network for it.
"Most network engineers still tend to think of video as size -- that a video call requires 200 times the capacity of a voice call," Jouret says. "But video is an experiential technology. You don't read it, you watch it. Video is like the acid test of the quality of your infrastructure."
Just like VoIP forced network engineers to build out low-latency, low-jitter networks, video will press the network even more. "If the video delivery starts losing more than a few packets, users will know, the call will drop," Jouret says.
Plus, as video works its way into the enterprise, its business uses expand. For instance, retailers have begun to explore video as a way to improve customer service. A system that may have begun as a surveillance system can be examined for trends, such as patterns in shopper traffic. Or it can be outfitted with facial recognition software to help store employees recognize when a store's most frequent customers enter the shop, says IDC storage and cloud analyst Rick Villars. Uses like this require that a lot of historical video is stored and then viewed and reviewed to find patterns.
Regular access of much-viewed video, particularly videos that will be viewed by many people at once, requires storing not just one copy of it, but multiple copies. These would be cached in some form of content distribution network, be it Cisco's new offering or another, as close to the device as is practical, so that the network can deliver the video without pause.
Jouret's answer is, naturally, to invest in Cisco's Enterprise Content Delivery System, announced last week. ECDS is part of the company's MediaNet enterprise video architecture. ECDS consists of software that can be run on one of two new appliances, the Media Delivery Engine 3100 and 1100, or on a virtual blade on a Cisco WAAS (Wide Area Application Services) platform. Jouret is touting some of Cisco's related video management products, such as MediaTrace, which allows IT professionals to troubleshoot video performance issues.
But besides storing and delivering video, Jouret's team is also developing new features to solve the problem of identifying content in the video. An upcoming product feature for its Show and Share Webcasting video product features automatic tagging. It listens to the video and automatically applies keywords. This can be done to tag an enterprise's legacy video library. It has a feature called speaker identification in which it automatically identifies all the speakers in a video, regardless of the language they are using. By searching on a speaker, an entire video will be reduced to just the portions where the selected speaker is talking. (See video below.)
Cisco believes the deadline for figuring out a video strategy for your network is fast approaching. "In three years, 91% of all network traffic on the Internet will be video. That means file transfers, bit torrent and other data will be the remaining 9%. So over the lifetime of your employees, they will work with video as the dominant data type. It's going to shape networks. It's going to make networks bigger and it's going to make networks smarter," Jouret says.
Julie Bort is the editor of Network World's Cisco Subnet community. She also writes the Odds and Ends blog for Cisco Subnet, the Microsoft Update blog for Microsoft Subnet and Source Seeker for the Open Source Subnet community sites. Follow Bort on Twitter @Julie188.