Monday, January 30, 2006

The Missing Link in QoS

I just can’t seem to get enough of this topic lately. I was reading the Stream of Consciousness blog wax on about how there are already tiered levels of service for Internet access. The tiers mentioned in this blog are for connecting to the Internet backbone itself. It does nothing to specify a level of service across many different backbones. The answer to improve quality of service is not to throw bandwidth at the problem, but manage the service through the standards we have already developed in the IETF like DiffServ. Consumers and application developers will always find new ways to consume the bandwidth. So you end up with a never ending spiral of adding and consuming bandwidth which is no way to manage a network.

What makes the Internet so great is that it is a collection of networks connecting the world together Metcalf’s Law comes true. The Internet with all its connected networks has contributed greatly to the last two decades of innovation in industry. Now that we have come to depend on it for daily living, we need to be able to manage the services that flow over it. BT is a leader in offering QoS for retail services and charge on a pay as you go basis. Other European providers do the same. North American consumers should have the same choice.

If application providers understood the benefit to managing QoS over the entire connection, then they could control their customer’s experience and manage their costs better. Google can leave their search engine service as best effort because it does not matter if it takes a few extra milliseconds for a result to come back. Vonage, on the other hand, would benefit by prioritizing its packets as they travel through the network. With deterministic latency and jitter, their voice service will have higher MOS and less service calls. Costs will decrease and customer satisfaction will increase so they can truly compete with POTS. Application/service providers that took advantage of guaranteed QoS would most likely pass the added cost on to the consumer receiving a double bump by increasing revenue and decreasing support costs.

What is missing is consumer control of QoS. If Yahoo!, Google, MSN, Vonage, Skype each gave their service the highest priority, the result to the user would be no different than the best-effort service they get today. The user needs some control over service prioritization. Manufacturers need to include QoS management tools in their CPE. D-Link is one such company that is developing broadband home routers and integrated access devices with QoS management tools. More are jumping on the bandwagon like Jungo, a popular CPE software application developer.

Why shouldn’t consumers be able to purchase these devices and have them work with their Internet access services? Smart QoS enabled CPE would automatically detect applications that need higher priority or have a higher QoS marking and appropriately balance the packet queues. The majority of users would not have to even know anything other than time sensitive applications like voice and video work great. Consumers would be more apt to stream video from Netflix or sign up for VoIP service. I see implementation of QoS on consumer Internet access as a boon to the industry.

Tags: VoIP, RBOC, Google, Vonage, D-Link, QoS