On Thu, Oct 30, 2008 at 10:13 PM, Patrick W. Gilmore <[EMAIL PROTECTED]> wrote: > On Oct 31, 2008, at 1:05 AM, vijay gill wrote: >> >> On Thu, Oct 30, 2008 at 9:41 PM, Patrick W. Gilmore <[EMAIL PROTECTED]> >> wrote: >>> >>> On Oct 30, 2008, at 10:19 PM, vijay gill wrote: >>> >>>> This is probably going to be a somewhat unpopular opinion, mostly >>>> because people cannot figure out their COGS. If you can get transit >>>> for cheaper than your COGS, you are better off buying transit There are >>>> some small arguments to be made for latency and >>>> 'cheap/free' peering if you are already buying transit at an exchange >>>> and your port/xconn fee is cheaper than your capital/opex for the >>>> amount of traffic you peer off. >>> >>> One of us is confused. >> >> precisely. > > Well, I could also be confused, which would make two of us. But I will > agree with you here and say precisely one. > > >>> Transit is _part_ of COGS, at least for most of the group reading this >>> list. >>> Finding transit "cheaper than your COGS" just means cheaper than you get >>> it >>> now. And that in no way way means you should dump peering. What if >>> peering >>> is cheaper than transit? >> >> Cost of transport, opex and capital to build out to a peering point, >> ports for interconnect, vs the expected money saved by peering away >> sufficient traffic is the analysis that will inform your strategy. >> This is why I said if you are already at a place where you are buying >> transit, it probably worth it to peer with the folks locally. > > None of that is in question. However, you also said: "If you can get > transit for cheaper than your COGS, you are better off buying transit and > not peering." So either you were confused, or .. well, let's be generous > and say you were confused. :-) > > I'm glad you have cleared your confusion.
Yeah, I was using COGs as a shorthand for cost to build out to peering locations, I should have really further broken it down into specific cases. /vijay > > -- > TTFN, > patrick > > >