[160333] in North American Network Operators' Group

home help back first fref pref prev next nref lref last post

Re: Global caches

daemon@ATHENA.MIT.EDU (Jeff Richmond)
Mon Feb 4 09:59:30 2013

From: Jeff Richmond <jeff.richmond@gmail.com>
In-Reply-To: <20130204145057.GD6663@virtual.bogons.net>
Date: Mon, 4 Feb 2013 06:59:16 -0800
To: Simon Lockhart <simon@slimey.org>
Cc: "nanog@nanog.org" <nanog@nanog.org>,
 Kyle Camilleri <Kyle.Camilleri@melitaplc.com>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org

While I would agree with that, having peering helps but certainly =
doesn't replace a localized CDN. Certainly better than nothing though. =
It also of course depends on the size of your network. If you are paying =
to carry that traffic (leased backhaul, etc.) from your peering point to =
your customers, you are still paying the same amount to deliver that =
content to your users (excluding any transit savings if moving from =
transit to get that CDN content). That is where an on-net CDN really =
saves you significantly as you can bury it deep into your network. I =
can't speak specifics here but I can tell you that the CDNs we have are =
filled at off-peak, so it really does become a win-win from a technical =
perspective (business case and politics are a completely different =
conversation though).=20

-Jeff

On Feb 4, 2013, at 6:50 AM, Simon Lockhart <simon@slimey.org> wrote:

> On Mon Feb 04, 2013 at 02:03:54PM +0000, Kyle Camilleri wrote:
>> Does anybody know of any other CDN providers that offer similar =
caches?
>=20
> Most CDN providers also provide free access to "super node" caches at =
major
> datacentres and peering points - depending on where you are located, =
which
> datacentres you're in, and what your network looks like, you may find =
that it's
> cheaper for you to interconnect with the CDNs within a datacentre =
(particularly
> if you can do it via an IX), than the provide space and power for CDN =
nodes=20
> within your own network.
>=20
> Simon
>=20



home help back first fref pref prev next nref lref last post