[j-nsp] LACP/LAG

Graham Brown juniper-nsp at grahambrown.info
Thu Oct 17 17:40:37 EDT 2013


LACP is great until you hit a box that is busy and doesn't offload this to
the line card - for busy switches and firewalls, use periodic-slow instead
of fast - I've had instances of EX and SRX that can't keep up with
periodic-fast and the LAG ends up being torn down during commits.

Bear in mind that this is only on busy boxes and we deploy LACP wherever
possible...

Just my $0.02


On 18 October 2013 10:34, Chris Kawchuk <juniperdude at gmail.com> wrote:

> I sometimes use LACP as well as a "poor man's BFD"; in the case of "the
> lights are on, but nobody's home" syndrome.
>
> aka a situation where the physical link(s) may be up, but the control
> plane functions are dead at the far end. Without LACP control packets, you
> may inadvertently start trying to send traffic down a link where the other
> end isn't actually functional yet. That's a definite case, albeit for a
> single-link LACP.
>
> If you can turn it on, and both sides support it, then I suggest using it;
> I haven't seen any harm IMHO.
>
> - CK.
>
>
> On 18/10/2013, at 8:00 AM, Keith <kwoody at citywest.ca> wrote:
>
> > Both sides came up on the MX and it looks ok. Am I going to get bitten
> in the ass
> > at some point for not running LACP?
>
>
> _______________________________________________
> juniper-nsp mailing list juniper-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>



-- 
Graham Brown
Twitter - @mountainrescuer <https://twitter.com/#!/mountainrescuer>
LinkedIn <http://www.linkedin.com/in/grahamcbrown>


More information about the juniper-nsp mailing list