[190103] in North American Network Operators' Group
Re: NANOG67 - Tipping point of community and sponsor bashing?
daemon@ATHENA.MIT.EDU (Mikael Abrahamsson)
Wed Jun 15 14:36:41 2016
X-Original-To: nanog@nanog.org
Date: Wed, 15 Jun 2016 20:34:36 +0200 (CEST)
From: Mikael Abrahamsson <swmike@swm.pp.se>
To: Sander Steffann <sander@steffann.nl>
In-Reply-To: <0DA9FCB4-32B9-4271-A79B-BFD1EABD5CEA@steffann.nl>
Cc: NANOG list <nanog@nanog.org>
Errors-To: nanog-bounces@nanog.org
On Wed, 15 Jun 2016, Sander Steffann wrote:
> I want to go to an IXP being a nice simple ethernet switch. Add some
> nice graphs and a route server, and we're done. Redundancy is a separate
> switch :)
So how should the larger distributed IXPs solve this? Provide optical DWDM
transport? Dark fiber? What if there are no chassis on the market large
enough to accomodate creating a single switch for all customers to connect
to? Have multiple L2 domains and require people to connect to multiple
switches? Even bumping people off of an existing switch when that is full,
to move some traffic over to another new switch chassis?
What about buffer requirements? If you want a buffered switch, it
increases capex and lowers number of switch-models that can be used.
Microbuffered switches may lose packets in microbursts when ports are
being run (near) full.
--
Mikael Abrahamsson email: swmike@swm.pp.se
