Video driving carriers to rethink their network architecture choices

Michael Kennedy, ACG Research

Michael Kennedy, Principal Analyst, ACG Research

It is not new information that video makes up most of total consumer IP traffic and is quickly over taking mobile data traffic. The types of equipment vendors are developing and how they are responding to the video challenge are what's new.

IP-based video traffic behaves quite differently than voice over IP (VoIP) data traffic (Web browsing and email).  Consumers' perceptions of video service and their willingness to pay also are quite different than those for voice telephony and Internet access services. These differences make it much more challenging to successfully provide IP video than it was to use IP to provide legacy voice and data services.

It is obvious that video content has much higher information density than voice, one-page-at-a-time Web browsing and email. Therefore, much greater bandwidth is needed to deliver video. However, bandwidth is not the only difference between video and other IP services. Video is flow based which means that transactions between video servers and individual subscribers must be set up and maintained with consistent quality for the duration of a video session. Voice also is flow based, but its bandwidth requirements are two orders of magnitude lower and its typical call duration is three minutes versus 20 to 90 minutes for video sessions. Also, video has been traditionally broadcast whether delivered over the air or on cable systems. Today, cable operators are aggressively promoting video on demand service which by definition is unicast, as is nearly all Internet video. Each of these differences between IP video and legacy services requires more network resources to meet subscribers' quality of experience expectations.

...there is a direct relationship between the network's capabilities and the price the subscriber is willing to pay for service. No such relationship exists for video services.

Consumers' perception of what video service is and their willingness to pay for it also is formed differently and valued differently than voice telephony or broadband Internet access. The most important difference is that telephony and Internet access service have always been offered as communications services. Telephone calls were traditionally priced on a distance and message-minute basis. Even today mobile voice services are packaged in minutes of use bundles. Internet access service also is marketed as a communications service with service tied to maximum upload and download data rates.  In both cases there is a direct relationship between the network's capabilities and the price the subscriber is willing to pay for service.  No such relationship exists for video services. Consumers value video service based upon the attractiveness and scope of its video assets not on downstream and upstream data rates or on minutes of use.  Consumers' willingness to pay is based upon their experience with substitute video delivery vehicles including broadcast TV, cable TV, and disk rentals. These alternatives put a ceiling on what consumers are willing to pay for IP video services.

The bottom line is that IP video requires a lot more resources than traditional voice, Web and email services and consumers' willingness to pay is constrained by substitute offerings and consumers' view that the content is where the value is not the delivery vehicle.

Systems vendors' response to these technical and market imperatives include increasing the bandwidth of the access network by moving fiber closer to subscribers and moving to including GE (gigabit Ethernet) or even 10 GE backhaul and extending QoS (quality of service) out to individual services and subscribers. Aggregation networks are particularly pressured by the large increases in bandwidth caused by the move to unicast video services.  Multiple 10GE DWDM (dense wavelength division multiplexing) transport links with 100 GE links to the core network provide part of the solution.  One alternative to throwing bandwidth at the problem is to use cloud computing concepts to distribute video caches out onto the aggregation network so as to reduce its required bandwidth capacity. Packet Optical Transport Systems (P-OTS) also play a role here in more tightly integrating routing, switching, and transport capabilities so that a single network element provides only as much of each function as is required, thus, reducing total cost of ownership.

Advanced edge routing concepts also provide alternatives to provisioning more bandwidth. Edge routers can deliver both service level and subscriber level QoS so that no more bandwidth than necessary is transmitted on the aggregation network. Edge routers also are augmented with intelligent networking capabilities (DPI, deep packet inspection) to better deliver application specific QoS and enforce service differentiation policies.

Another strategy for delivering QoS and other policy management capabilities all the way into the access network while substantially reducing cost is to use a centralized routing and switching control protocol so as to take the cost of advanced routing processors out of the access and aggregation network nodes. Various approaches employ PBB-TE (provider backbone bridging-traffic engineering), MPLS-TP (multiprotocol label switching-transport profile), and the Ethernet tag switching protocols. The economic benefit comes from employing much lower cost Carrier Ethernet switches on the aggregation network rather than fully equipped Carrier Ethernet Switch Routers. The Ethernet switches can be standalone systems or packaged as Ethernet line cards on optical transport equipment.

Finally, core networks are being simultaneously increased in bandwidth capacity through the introduction of 100 GE router ports and cost reduced through the use of P-OTS (which reduces the cost of aggregating core network links and switching transit traffic. The cost savings is realized because transport ports are substantially lower cost than core network router ports.

Emerging network architectures and designs are being shaped by the need to meet the demanding performance requirements of video traffic while lowering unit cost points to make IP video delivery economically viable. The resulting networks will have substantially larger bandwidth capacity than existing networks and support intelligent service delivery to all access network end-points.

Michael Kennedy is a regular FierceTelecom columnist and is Principal Analyst at ACG Research. He can be reached at [email protected]