40G reforming

Tim Jackson jackson.tim at gmail.com
Tue Feb 6 02:11:14 UTC 2018


I'm pretty sure that this is only available on 7150S which is FM6000, not
broadcom at all.



On Feb 5, 2018 8:00 PM, "Ryan, Spencer" <sryan at arbor.net> wrote:

You don’t use 40G modules at all. Just 4 x 10G SFP+.

The Broadcom trident chip is configured at the MAC layer for 40G, so it’s
identical to a real 40G port inside.

Some more reading:

https://www.arista.com/assets/data/pdf/Whitepapers/
AgilePorts_over_DWDM_Final.pdf


Spencer Ryan | Senior Systems Administrator | sryan at arbor.net<mailto:[email protected]
arbor.net>
Arbor Networks | The security division of NETSCOUT
+1.734.794.5033 (d) | +1.734.846.2053 (m)
www.arbornetworks.com<http://www.arbornetworks.com/>



From: Hunter Fuller [mailto:hf0002+nanog at uah.edu]
Sent: Monday, February 5, 2018 2:57 PM
To: Ryan, Spencer <sryan at arbor.net>
Cc: Marian Ďurkovič <md at bts.sk>; Baldur Norddahl <baldur.norddahl at gmail.com>;
nanog at nanog.org
Subject: Re: 40G reforming

I suspect that implies that you can just take a 40Gbase-SR4 module and
break it out into individual "10G" multi-mode pairs for DWDM use. Has
anyone tried this? I'm also very interested in using that strategy.

On Mon, Feb 5, 2018 at 1:36 PM Ryan, Spencer <sryan at arbor.net<mailto:[email protected]
arbor.net>> wrote:
Indeed. Arista does (did?) make at least one platform where you can do this.

-----Original Message-----
From: NANOG [mailto:nanog-bounces at nanog.org<mailto:nanog-bounces at nanog.org>]
On Behalf Of Marian Durkovic
Sent: Monday, February 5, 2018 2:33 PM
To: Baldur Norddahl <baldur.norddahl at gmail.com<mailto:baldur.norddahl at gmail.
com>>
Cc: nanog at nanog.org<mailto:nanog at nanog.org>
Subject: Re: 40G reforming

Many switches based on BCM Trident ASIC allow you to configure 4 consecutive
SFP+ ports as 40G link (not LACP, but using real hardware 40G framing).
In such case, you can plug 4 DWDM SFP+ modules directly into the switch,
without the need for any reformer.

   M.

On Mon, 5 Feb 2018 20:03:33 +0100, Baldur Norddahl wrote
> I may need to clarify that I do not want to break the port into 4x10G
> as such. To the switch this will be an ordinary 40G link to another
> switch far away.
>
> I want to take advantage of the fact that 40G is transported as four
> individual streams. Each of the four streams are to be converted from
> 850 nm to a 1550 DWDM channel (one channel per stream). And the
> reverse at the other end of the link.
>
> The point of doing this is that 40G DWDM modules are not generally
> available and neither are 80 km modules.
>
> I need a true 40G channel so 4x10G LACP is not an option here. For the
> same reason I am unable to accept a solution that splits the 40G port
> into 4x10G and then perhaps recombines using LACP. Instead I am
> looking at an optical solution that is invisible to the switch hardware.
>
> The only doubt I have about the proposed solution is whether the frame
> format of the 10G substreams is somehow incompatible with what goes on
> in the reformer. As I understand these reformers they are little more
> than two SFP(+) modules connected back to back. And therefore it
> should not matter that the frame format may be different.
>
> Regards
>
> Baldur
>
> Den 5. feb. 2018 7.20 PM skrev "Paul Zugnoni" <pz at wish.com<mailto:pz at wish.
com>>:
>
> Whether a 40G port can be broken into 4x10G is dependent on the
> router/switch hardware and the optic you use. Good news is that most
> 40G ports are capable of being broken out into 4x10G, since a 40G port
> is usually operating as 4x10G internally anyway to the ASIC. The QSFP
> you'll need would be a 40G-SR4 for MTP/Multimode or 40G-LR4 for
> MTP/Singlemode (or a lower power, less expensive equivalent). This is
> a pretty common use of 40G ports. All 4 10G ports would then be at
> 850nm or 1310nm, which you can then plug into any 10G SR or LR ports.
>
> What router or switch platform is driving the 40G?
>
> Paul Z
>
> On Mon, Feb 5, 2018 at 7:57 AM, Baldur Norddahl
> <baldur.norddahl at gmail.com<mailto:baldur.norddahl at gmail.com>>
> wrote:
>
> > Hello
> >
> > Is it possible to reform a 40G signal as individual 10G links?
> >
> > The idea is to use a 40G QSFP multimode MTP module such as
> > https://urldefense.proofpoint.com/v2/url?u=https-3A__www.fs.com_prod
> > ucts_44058.html&d=DwIDaQ&c=Hlvprqonr5LuCN9TN65xNw&r=Iw8ah1pcqZhOErIj
> > aFRfuA&m=wWoshgttJT0E6q6-qJzP_ZcIrEz_EP88taPCbvAiK2Y&s=_rJfOmyDlGmPG
> > C6M5FbhQ1V8_mho1OCpkcuYRNlaOvA&e=. Then connect it using a MTP
> > breakout cable such as
> > https://urldefense.proofpoint.com/v2/url?u=https-3A__www.fs.com_prod
> > ucts_68049.html&d=DwIDaQ&c=Hlvprqonr5LuCN9TN65xNw&r=
Iw8ah1pcqZhOErIjaFRfuA&m=wWoshgttJT0E6q6-qJzP_ZcIrEz_EP88taPCbvAiK2Y&s=
Cz0mCyM3dtcHoZ7lGy7uyroI_Y7AwmKXdnYNFIF0rPI&e= to get four dual fiber
connectors. These are then connected to four 10G SFP+ multimode modules
such as https://urldefense.proofpoint.com/v2/url?u=https-3A__www.fs.
com_products_11589.html&d=DwIDaQ&c=Hlvprqonr5LuCN9TN65xNw&r=
Iw8ah1pcqZhOErIjaFRfuA&m=wWoshgttJT0E6q6-qJzP_ZcIrEz_EP88taPCbvAiK2Y&s=l-
9OAiUxeydRJCJc7d1kTKPVSkwQlkV4xkZFlbFxyRs&e=. The reformer could be
https://urldefense.proofpoint.com/v2/url?u=https-3A__www.fs.
com_products_43721.html&d=DwIDaQ&c=Hlvprqonr5LuCN9TN65xNw&r=
Iw8ah1pcqZhOErIjaFRfuA&m=wWoshgttJT0E6q6-qJzP_ZcIrEz_
EP88taPCbvAiK2Y&s=NwCHiC_boNNs7zCOgJFRZ5nmZOVEPBovGYNTtdQ_pCE&e=. And
finally the reformed signal can be transported using anything including
DWDM modules such as https://urldefense.proofpoint.
com/v2/url?u=https-3A__www.fs.com_products_44058.html&d=DwIDaQ&c=
Hlvprqonr5LuCN9TN65xNw&r=Iw8ah1pcqZhOErIjaFRfuA&m=
wWoshgttJT0E6q6-qJzP_ZcIrEz_EP88taPCbvAiK2Y&s=_rJfOmyDlGmPGC6M5FbhQ1V8_
mho1OCpkcuYRNlaOvA&e=.
> >
> > Just using fs.com<https://urldefense.proofpoint.com/v2/url?u=http-
3A__fs.com&d=DwMFaQ&c=Hlvprqonr5LuCN9TN65xNw&r=Iw8ah1pcqZhOErIjaFRfuA&m=
s0btKmr0Z2yc5j4taNFqdO13c28jnsjkla2DQeNIkIY&s=FSMpOpeWl32KxNR-rraOUqpV8n8eP-
nhn4jTT8N3mWQ&e=> as a reference to the kind of equipment I am
> > talking about. Many other vendors offer simelar products.
> >
> > The motivation for doing this is to get access to the many options
> > that are available for 10G optics but not possible with 40G.
> >
> > Regards,
> >
> > Baldur
> >
> >
--

--
Hunter Fuller
Network Engineer
VBH Annex B-5
+1 256 824 5331

Office of Information Technology
The University of Alabama in Huntsville
Systems and Infrastructure



More information about the NANOG mailing list