On 19 Feb 2014, at 18:29, Darrel Lewis (darlewis) <[email protected]> wrote:

> 
> On Feb 19, 2014, at 9:18 AM, Dino Farinacci <[email protected]> wrote:
> 
>>> 1.  if you store the cache for next reboot, you will not experience
>>> the miss storm when the traffic will come back to you.
>>> 
>>> 2.  if you shutdown an ITR, packets are forwarded to another ITR and
>>> there is a miss storm as long as the prefixes in the backup ITR do not
>>> cover those that where in the "down" ITR.
>> 
>> LISP promotes multi-homing for inbound traffic to a LISP site, as well as 
>> outbound packets from a LISP site. You deploy LISP so you can get 
>> active-active multihoming.
>> 
>> So this storm will typically be academic.
> 
> +1 
> 
> I've always thought that the mapping system will have to have capacity to 
> deal with broken clients, random LIG bots, and DoS attacks (just like, say, 
> DNS does).  This leads me to believe that over-engineering the management of 
> sending map-requests on a given ITR will likely end up creating more 
> complexity, and therefor fragility, in the implementation.
> 

That's a point we can discuss, but are you ready to accept packet loss?
At the end, a miss = a packet drop...

Damien Saucez

> 
> -Darrel

_______________________________________________
lisp mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/lisp

Reply via email to