On Aug 27, 2010, at 7:57 PM, Peter Sahlstrom wrote: > This may be a stupid question, but I can't come up with an answer, so here > goes. > > Are there really applications that can't be supported by the > best-effort internet? Wasn't one of the foundational principles of > ARPANET development that the network would be treated as unreliable, > and that the endpoints would compensate for this?
It boils down to what you mean by supported. Highly interactive applications will benefit to the point where best effort with any sort of congestion or other traffic impact makes a noticeable dent. Two boilerplace scenarios where this is not preferred would be gaming (noone likes it when the raid dies because the healer is lagging) and financial trading applications. These are two cases where we see a lot of effort being put into either avoiding best effort networks altoghether, or trying very, very hard to mitigate the effects as much as possible. > I just find it odd that, for example, VoIP grew up as a stable > alternative technology built on top of an unreliable network, and now > that phone companies are starting using VoIP in place of POTS, they're > arguing that they need a reliable network to run it on. There's voip and voip. Voip on the consumer side ment cheap minutes. Voip on the telco side ment not having different standards and transports every bloody step of the way. Different drivers and, in a way, different needs. > -Peter Kriss
