NTT's Video Tunnels

9:30 AM -- NTT Group (NYSE: NTT) says it has developed a new transmission system that enables the creation of on-demand, automated video circuits across a packet-optical network. The Japanese operator says the system includes in-built video quality monitoring, and has been developed specifically for the delivery of bandwidth-hungry wideband video. (See NTT Tests Video Transport.)

The development sounds like some kind of video transport nirvana. "With this advancement, users can achieve steady transmission of high-density video such as HDTV by acquiring a path in the network whenever it is needed, without allocating additional network resources or re-optimizing the provisioned network resources," states NTT.

The carrier says it has tested the system with Japan's National Institute of Information and Communications Technology (NICT), and that it intends to show off its "synergy of video applications and IP-optical networking" at the upcoming SC09 event in Portland, Oregon (November 14-20).

NTT says current systems for the transport of uncompressed video, which requires "bandwidth of more than a gigabit per second," aren't capable of real-time network asset provisioning, and that the level of monitoring needed for video "requires special equipment that is too expensive to distribute to many points in the network."

The Japanese carrier says it's "planning further enhancements to network and video collaboration that will allow greater functionality, such as adaptive control, so that it can be employed by broadcasters and other high-end users. Furthermore, our aim is for this technology to eventually be used in many other arenas that can benefit from high-resolution video, including telemedicine and e-learning."

The point here, it seems, is that having technologies that can provide lots of bandwidth isn't enough -- the real pain points are in the network and service operations, in this case the provisioning and service assurance. And as long as operators like NTT feel they can't source the capabilities they need from the vendor community, they're going to crack on and develop them in their own labs, something at which NTT appears to be rather good: It recently won a Light Reading Top Pick award, in the IP Applications category, for its ongoing work on IMS (IP Multimedia Subsystem). (See Light Reading Reveals Its 2009 Top Picks.)

— Ray Le Maistre, International News Editor, Light Reading

joferrei 12/5/2012 | 3:52:48 PM
re: NTT's Video Tunnels

Interesting.. so NTT developed their own system for video transport? Any idea if this included also the equipment, or just the management system?

What may be their technology differentiation, compared with the at least somewhat related developments such as STS-X VCAT, Nx1.25Gbps ODUflex, OBS etc.?  And do they have intellectual property rights advantage, and how could that affect other service providers in competitive video IPTV etc. service markets?

What can NTT’s dynamic video transport technology do that the commercial vendor technologies (e.g Cisco, ALU etc.) cannot?

torivar 12/5/2012 | 3:52:44 PM
re: NTT's Video Tunnels

They developed their own video transport packet format awhile back.   That is their i-vista system, for which they developed the hardware and software.   It's meant to packetize and transport high rate (uncompressed) video traffic, with the packetization being near realtime.

My guess is this is just the management system   NTT has done a lot of research into multi-layer networks and have developed control-plane software for setting up optical wavelengths/circuits on-demand.   They did one of the first GMPLS signaled circuits from router to optical network to router with Juniper a few years ago.  

Internet2 has had similar constructs but it's driven by users selecting two points and creating on-demand circuits.  Not sure if the NTT system works similarly.

joferrei 12/5/2012 | 3:52:39 PM
re: NTT's Video Tunnels

As IP video type traffic is forecast to become the dominating form of network traffic, is there some indication that these dynamic optical wavelength/circuit switched technologies might become the dominating type of transport networks going forward?

And will the ownership of the enabling technologies / patents give the leading service providers a level of differentiation in the otherwise commoditizing network capacity services market place?


torivar 12/5/2012 | 3:52:39 PM
re: NTT's Video Tunnels

I think you'll see the move away from using routers for packet transport of broadcast and on-demand video onto packet optical systems or just optical systems, mainly due to port cost.   I don't know about the dynamic wavelength/circuit setup part, I think we are a ways away from seeing that become the norm.  

Building a broadcast video network on top of a L3 IP/MPLS network adds a ton of complexity to the network which probably doesn't need to be there, you just need to build a P2MP distribution tree.   As traffic levels get near 10Gb/s it doesn't make sense to add more 10Gb/s router ports when an optical transponder or packet optical port is 1/10th the cost.

I think the push to convergence over a common L3 IP/MPLS network is fading as traffic levels of different service types approach 10Gb which is realistically the largest link size most networks have today.   

joferrei 12/5/2012 | 3:52:35 PM
re: NTT's Video Tunnels

Obviously plain optical channel connections are lower cost/bit than any packet level processing or routing.. but don't we have for such optical vs. IP layer video transport the same dilemma as with any packet traffic delivery: What one saves on equipment port cost one more or less pays via needing more bandwidth due to lower port utilization, per


Or are the economics substantially different for video networking?

And what if video type traffic becomes the norm for next gen communications, replacing VoIP and plain data transfer--won't routed IP like ubiquitous connectivity then be needed for these 'video tunnels' as well, and how to then by pass IP routing?

torivar 12/5/2012 | 3:52:34 PM
re: NTT's Video Tunnels

It depends on the video application.  If you are talking about linear broadcast video in which the is a single root source and every end site gets the same copy of the video, that's a good candidate for router bypass.  In the end routers still process the data because almost all video delivery today is done over multicast IP.   Just the intermediate core routers are bypassed because the tree is pre-built and deterministic.  This if for high bandwidth distribution where it does not make sense to converge.  

Multipoint video delivery where any endpoint can both send and receive video can still be a candidate for router bypass.   Using a control plane like BGP-based multicast VPN (ng-MVPN) PE routers can signal LSPs on demand between sources and sinks, which is only one step away from being able to signal a wavelength.    That's a bandwidth decision more than anything else.   

With standards like ODUFlex able to signal a sub-LSP across OTN of Nx1.2Gb size your router can be connected to an OTN network via 10Gb,40Gb,or 100Gb and use GMPLS to signal an OTN LSP to directly connect two PE routers.  Instead of having a fixed 10Gb pipe to a core router with a fixed pipe to another router, etc.  

I don't know, as a packet/router guy I'm hoping they just make router ports cheaper so we can continue to use a single ubiquitous IP/MPLS core for all services, instead of all this grooming/bypass/etc.   We have been too slow to get to 40Gb or 100Gb router ports and the cost of them going to be too expensive when they do come out.  Why upgrade your DWDM system to handle 40G and 100G ports AND upgrade every router port as well.   Your core could just be one system...  

Sign In