re: [RRG] cache issues in LISP and CONS

[Scott Brim <swb@employees.org>]

On 22 Oct 2007 at 10:36 -0400, Noel Chiappa allegedly wrote:
> The issue is not who has what information, but rather the size of
> the cache one would need in a CDR (particularly a top-level CDR) for
> that cache to be effective.

I agree with your goal of not requiring a full cache, but the main
problem is that the aCAR has no idea what a CDR's operating
environment is, so how can the aCAR predict when "it would be good to
cache this"?  All the aCAR can tell the CDR is what *it* is
experiencing.  Telling the CDR when to cache will certainly be either
too early or too late -- unless of course you built knowlege of CDR
performance into the aCAR, which sounds like a horrible idea.

Let's go with your assumption that the CDR has too many "chaff"
queries, i.e. there are so many one-off mapping requests that the
"occasional" ones (not rare, but not frequent either) are pushed out
of the cache.  I guess I would accept something in the map-reply that
wouldn't "advise" the CDR on what to do, but might indicate the aCAR's
experienced rate of queries ... but caching by CDRs *other* than the
one in question will reduce that rate to be the same as the "rare"
rate.  At a system level, an aCAR reports how the behavior of *all*
CDRs is affecting it, and a *single* CDR tries to figure out how to
behave in response?

So again, I agree with the goal, but I don't like the mechanism.

This would be a good thing to simulate.

swb


--
to unsubscribe send a message to rrg-request@psg.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg