[bind10-dev] more about cache (Re: cache effectiveness)

[JINMEI Tatuya / 神明達哉]

At Tue, 5 Mar 2013 08:25:46 +0100,
Michal 'vorner' Vaner <michal.vaner at nic.cz> wrote:

> I'd just like to point out, even when some kind of RPC would be slow, we still
> can use some kind of shared memory (written by the cache process only, but open
> read-only by the resolvers), which would still be better I believe than fight
> with threads.
> 
> But for the performance, I believe this model would be similar to threaded
> version with shared cache. Or is there some overhead I don't see?

The main concern was indeed the overhead of interprocess
communication, and I simply didn't know how much it would affect the
overall system performance.

In any case, I doubt the use of shared memory works as a silver
bullet.  We'll then just suffer from interprocess synchronization
problems instead of inter-thread ones.

---
JINMEI, Tatuya
Internet Systems Consortium, Inc.