Hello;
Wouldn't the transverse of the routing and/or IP filter tables be a good
cause not to do such things?
Warmest Regards
Steven Sim
James Carlson wrote:
Octave Orgeron writes:
I've used ndd to increase the default to support 1000 vip's before and
have not noticed any performance issues.
And you shouldn't -- unless you have applications that are affected by
high numbers of interfaces.
So it won't hurt your server.
"Maybe." The kernel and usual sorts of applications work just fine
like this.
It really just comes down to the application that will use the NIC. If
it can co-exist with other apps on the same NIC and not cause any
bottlenecks, you shouldn't have any issues.
That's not the problem.
There are some applications that use SIOCGLIFCONF and/or routing
sockets to keep track of the interfaces on the system. These
applications often build internal lists of interfaces for their own
purposes. Unfortunately, some of those applications have O(N^2) or
worse algorithms that iterate over those lists. If you have dozens or
hundreds of interfaces, that's probably not a problem. If you have
many thousands, you can shoot yourself in the foot.
(I don't think this is really a _good_ reason for the hard-coded 8192
limit. It's the only reason I know, though.)
Fujitsu Asia Pte. Ltd.
_____________________________________________________
This e-mail is confidential and may also be privileged. If you are not the intended recipient, please notify us immediately. You should not copy or use it for any purpose, nor disclose its contents to any other person.
Opinions, conclusions and other information in this message that do not relate
to the official business of my firm shall be understood as neither given nor
endorsed by it.
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
