[177738] in North American Network Operators' Group
Re: Cisco Nexus
daemon@ATHENA.MIT.EDU (George Herbert)
Mon Feb 2 20:03:16 2015
X-Original-To: nanog@nanog.org
From: George Herbert <george.herbert@gmail.com>
In-Reply-To: <20150203002415.GB23129@radiological.warningg.com>
Date: Mon, 2 Feb 2015 17:03:06 -0800
To: Brandon Ewing <nicotine@warningg.com>
Cc: "nanog@nanog.org" <nanog@nanog.org>
Errors-To: nanog-bounces@nanog.org
> Brandon Ewing <nicotine@warningg.com> wrote:
>=20
>> David Bass wrote:
>> The n2k ToR is not a great design for user or storage interfaces if most o=
f your traffic is east/west. It is great as a low cost ilo/drac/choose your=
oob port, or if most of your traffic is north/south. Biggest thing to reme=
mber is that it is not a switch, and has limitations such as not connecting o=
ther switches to it. Like anything else you have to understand the product s=
o that you don't engineer something that it wasn't designed to do.
>=20
> And remember -- The Nexus 2K performs absolutely ZERO local switching -- a=
ll
> frames received from client ports are just copied to the upstream device, s=
o
> it can handle the frame/packet forwarding logic. =
=20
What this really does is force you to consider how much of your East-West is=
rack-local, versus off rack.
Rack-local-heavy hurts as badly as off rack, with FEX.
If you want to / can localize E/W tighter than that then you want real TOR s=
witching. If the average E-W is cross rack then the FEX are performance equ=
ivalent. For random distributions this comes at a few racks. For intention=
al distributions it's probably better to TOR switch from day one.
George William Herbert
Sent from my iPhone=
