At 21:18 06.05.2005, Neil Harris wrote:
[...]
You might also want to consider that Wikipedia is, on
balance, a large
net source of network traffic, and that peering agreements make most
sense for networks which have balanced traffic in and out (and, for
the same reason, transit gets cheaper for balanced flows, even if you
cannot negotiate settlement-free peering). Thus, network providers
with net inflows may well be able to make substantial cost _savings_
by giving Wikipedia free bandwidth, power and rackspace.
Wikipedia is rapidly becoming big enough to become a player in the
peering game, or at least should be aware of its advantage to others
in doing so. I presume this has not escaped the attention of some of
our potential benefactors...
-- Neil
Gerard Meijssen wrote:
Georg Chytil wrote:
While this was generally true in the past, your mileage will vary
nowadays : the costs for hardware and operations are
significant as compared to transit costs, the benefactor has no
real/immediate monetary benefit from hosting a wiki site IMHO.
Plus - you need to be in the Gigabit range to convince a tier 1
provider of establishing peering(s) with you.
Hoi,
The Dutch and the Belgian organisations are both in the Gigabit range.
Yahoo and this other "mystery" organisation are in that bracket as well.
Thanks,
GerardM
#g
Indeed. Consider the rate at which Wikipedia is growing, and it looks
likely that it will have flows in the multi-Gigabits range soon. At that
point, the logic of peering tips over the other way. Remember that
peering isn't a magic alternative to paying for transit bandwidth:
no-one will peer with you unless it is in their economic interest.
However, Tier1s have a strong economic incentive to peer with other
high-flow networks at points where both traffic and economic incentives
are balanced. Otherwise, they are just wasting money on unnecessarily
hauling traffic over long distances on their backbone networks, at great
expense to themselves.
-- Neil