Re: [lisp] Fwd: [rrg] [GROW] Operational experience with cache basedmapping ID

> |When the working set exceeds the capacity, you install a
> |single course
> |entry and use network diameter and network hierarchy to scale "the
> |system wide cache".
>
> In other words, punt to the next level of cache, which then has an  
> even
> larger working set.  How does that help?  It degenerates into a  
> centralized
> full map, which clearly doesn't work.

Tony, you know how hierarchy works. You follow coarse prefixes as you  
go up and more specifics as you go down. See draft-meyer-lisp- 
cons-04.txt for details. And refer to the VA stuff Paul is working on.

> |Choose your tradeoff of cache size versus data-
> |path stretch. We might have to choose both in different parts of the
> |network.
> |
> |Rather than anyone saying what is wrong with caching, present an
> |alternative which is better.
>
> First, we have to be free to say what is wrong.  Bringing up bugs in  
> a free
> and open way is necessarily part of the process.

Right, and we don't want that to stop. But would like some  
alternatives as well.

> Second, there have been numerous alternatives discussed: full map,  
> DHT, and
> hybrids.

Right, and I think many of these are misunderstood because they are  
not brought to the level of detail to find the bugs.

Dino

_______________________________________________
lisp mailing list
lisp@ietf.org
https://www.ietf.org/mailman/listinfo/lisp