Wietse:
> What is the latency for looking up information that is NOT in the
> memcache? If it is 10 millseconds, then postscreen can handle only
> 100 connections per second, and it becomes a performance bottleneck.
bi...@dev-ops.pl:
> I have no idea what is the latency for lookup not cached information.
Perhaps you should find out how much time this takes, so that you
won't be disappointed when (not if) there is a burst of traffic to
your server.
> But I think I don't really understand how it's done. Why postscreen cant
> handle more than 100 connections per seconds with usage of shared
> memcache?
If it take 10 milliseconds for postscreen to find out that an SMTP
client has NO WHITELIST CACHE ENTRY, then postscreen can do that
only 100 times per second.
You are talking about the best-case performance where postscreen
looks up information that is already cached.
But focusing on the best-case performance is a mistake.
You need to plan for the worst-case scenario where one of your users
is the victim of a joe-job, and many "new" systems connect to your
server. Or someone starts a spam campaign with a large number of
"new" zombies. Those systems will have NO WHITELIST CACHE ENTRY.
Wietse
