Irwin Lazar <ILazar@tbg.com> writes:

From what I've seen, there isn't a simple answer. In places where bandwidth is exorbantantly expensive (such as outside the United States), simply over provisioning isn't an acceptable answer. However, in some places in the U.S. & Europe, over-provisioning may certainly make more sense.

I think we're running into some semantic issues here. What you call "overprovisioning", I call "not underprovisioning" or "accurately predicting growth". There has never been a shortage of people willing to market something-for-nothing schemes to the gullible. Over the years, there have been perpetual motion machines, gizmos that you put in the fuel line to "magnetize" your gasoline and make your Eldorado get 50mpg, and end-to-end QoS schemes. If you were around for Peter Lothberg's presentation at nanog-last, you know what I'm driving at here.

In our area, we're also seeing a lot of pushback against the continued tearing up of streets to lay additional fiber,

If you put a reasonable amount of conduit in the first time (conduit is *cheap*), there is no reason to tear up the streets again to put in more fiber. Most of the street-re-tearing-up that I've seen done has been additional vendors, and no QoS mechanism that I've seen addresses that little problem. If you put in a reasonable amount of glass, and you're running low on strands, metro area sparse-wdm schemes are getting cheaper all the time, and presumably you can support their deployment because you have high demand (else why are you running out of fiber?). Notice that these are *quantity* of service mechanisms...

so QoS may become the only option to meet required service levels.

Or you may have to face the fact that it is impossible to provide the desired level of service at a price that people are willing to pay, in your particular locality. For instance, QoS schemes or no, I don't think any of the major (or minor for that matter) players have plans to put a datacenter on Pitcairn Island. ---rob