On Mar 20, 2013, at 6:47 PM, Noel Chiappa wrote:

>> From: Ronald Bonica <[email protected]>
> 
>> So far in our discussion, we have mention a few possible mitigations.
> 
> Forgive me if I'm being repetitive, but is the following on the list (sorry,
> a bit distracted at the moment, not keeping up with the thread):
> 
> - A fixed size cache, with a 'good' cache management algorithm.
> 
> Simple LRU would not be 'good', as if the cache is being thrashed fast
> enough, you might lose useful entries.
> 
> When this first came up a couple of years back (too lazy to look for it in
> the archives, sorry), I had suggested (IIRC) a two-level cache, where entries
> were only promoted to the second level when they'd been used a certain number
> of times; that way, only the first level could be thrashed by a DoS attack.
> 
> But in retrospect, that's probably more complex than it needs to be: an aged
> usage counter system would probably be good enough. Entries that got loaded
> as part of a DoS attack would presumably not be used as much as 'real'
> entries, and would be preferentially discarded.


Another useful piece of information would be _which EID instigated_ the map 
request.  If a host behind an ITR is under a randomly sourced syn flood, all 
the map-requests generated by that host could be treated as a lower priority 
(and aged out faster) than those cache entries instigated by other hosts.

There are, literally, dozens of techniques to optimize a cache that is under 
pressure from DoS attacks.

-Darrel

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

Reply via email to