Quality of Service vs. Experience
By John Parkinson
"Quality of service" is a pretty good way to set out a range of service level guarantees in a network. By defining traffic types (data, voice, video), performance requirements (bandwidth, latency, jitter and so on) and service characteristics (priority, price), carriers can try to ensure that the user experience is at or close to whatever expectations they (the user) will pay for.
QOS is based on the twin theories that a) not all traffic needs or deserves that same level of transport and b) prioritization and management is possible and desirable. In current networks, QOS underpins the ability to carry many traffic types simultaneously without undue performance issues and at attractive prices.
But in some circumstances, most notably on the public Internet, it's close to impossible to guarantee QOS. There is just too much unpredictability and volume of traffic, and not enough management capability in the network.
As the hype around the "cloud" model of infrastructure provisioning continues to grow, service providers have to find ways to finesse the "QOS on the Internet" problem. After all, a lot of their hoped for customers are going to use the public Internet to connect to the "cloud" and they are going to expect their connections to work and work well enough to do "the job."
Enter "Quality of Experience." The theory seems to be that a) users don't have very high expectations of the Internet experience and b) as long as it's "good enough," the actual quality doesn't matter. Because if you can't actually guarantee any particular level of quality, you shouldn't even try.
To their credit, a few service providers are rebelling against this overly cynical view and declaring that "good enough" is actually an admission of defeat.
I hope their views prevail, but I am not optimistic...I own a cell phone.
John Parkinson is CIO of TransUnion. To read his columns for CIO Insight, click here.