[j-nsp] Limit on interfaces in bundle

Edward Dore edward.dore at freethought-internet.co.uk
Thu Oct 29 08:57:11 EDT 2015

On 29 Oct 2015, at 12:49, Mark Tinka <mark.tinka at seacom.mu> wrote:

> On 29/Oct/15 14:22, Cydon Satyr wrote:
>> Oh wow.
>> Any real drawbacks to running something like 32x10Gbps LAG link in core
>> instead of higher bandwidth physical links? Just seems so unreal.
> Folk like AMS-IX have publicly acknowledged running 32x 10Gbps on their
> exchange point (albeit, with Brocade) right before 100Gbps ports became
> viable.
> I suppose the biggest issue will be how you hash equally across all of
> the links, especially if much of the traffic being carried is inherently
> Layer 2 (despite having a Layer 3 payload).
> Mark.
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp

I believe LINX were using 64x10G LAGs between the core nodes on the Juniper LAN in London before the upgrade to use smaller 100G bundles.

I guess the biggest benefit of using 100G interfaces instead of 10x10G is that you can support individual flows >10G.

Edward Dore
Freethought Internet
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 496 bytes
Desc: Message signed with OpenPGP using GPGMail
URL: <https://puck.nether.net/pipermail/juniper-nsp/attachments/20151029/4009d088/attachment-0001.sig>

More information about the juniper-nsp mailing list