[187956] in North American Network Operators' Group

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

Re: Duplex negotiation over 100Base fibre

daemon@ATHENA.MIT.EDU (Jared Mauch)
Mon Feb 29 13:01:12 2016

X-Original-To: nanog@nanog.org
From: Jared Mauch <jared@puck.nether.net>
In-Reply-To: <7024EEDA-5401-4470-B944-89DC33647DE1@lixfeld.ca>
Date: Mon, 29 Feb 2016 13:01:12 -0500
To: Jason Lixfeld <jason+nanog@lixfeld.ca>
Cc: NANOG <nanog@nanog.org>
Errors-To: nanog-bounces@nanog.org


> On Feb 29, 2016, at 11:07 AM, Jason Lixfeld <jason+nanog@lixfeld.ca> =
wrote:
>=20
> Hello,
>=20
> My understanding is that for 1G and 10G optical networks, there is no =
concept of half-duplex mode, but I=E2=80=99m unclear about half duplex =
in the 100M optical world.  Specifically, if I connect two 100Base-LX =
(or BX) transceivers together, is there a requirement for the =
controller(s on either side) to select between full or half duplex, =
either by static configuration, or auto-negotiation?


I can say even at 1G speeds, the fiber is just the medium, the =
underlying data stream can still be half-duplex or require =
autonegotation or the lack of it. I recall many years ago having to =
adjust settings for GSR <-> Foundry negotiation and how it would vary =
depending on the device, so if you changed from GSR -> 6500/7600, you =
may have to make changes to maintain link.

Having a standard device/switch/config will help you here.

Most media converters I=E2=80=99ve seen have a switch on the side that =
lets you set if it should advertise autonegotation, take the super-cheap =
MC220L as an example, it has a switch for this purpose.

After many years of hard coding things to gig/full, i=E2=80=99ve started =
to see the autonegotation work much better and require less custom =
coding of the config.  This doesn=E2=80=99t stop someone from plugging =
in really-old stuff, and we still see people connecting at 10/half or =
10/full when I really wonder what they are connecting.

- Jared=

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