Re: [Vserver] many ip addys

From: Chuck <chuck_at_sbbsnet.net>
Date: Mon 10 Apr 2006 - 21:45:29 BST
Message-Id: <200604101645.29549.chuck@sbbsnet.net>

On Monday 10 April 2006 02:10 pm, Herbert Poetzl wrote:

whoah... heh ok maybe i will implement 'plan B' for now and wait until this is
all settled in. those figures are a bit too much for us to use presently.

> On Thu, Apr 06, 2006 at 07:00:32AM -0400, Chuck wrote:
> > On Friday 21 October 2005 12:23 pm, Chuck wrote:
> > > On Friday 21 October 2005 12:12 pm, Herbert Poetzl wrote:
> > > > On Fri, Oct 21, 2005 at 06:58:38AM -0400, Chuck wrote:
> > > > > i forget.. was it ngnet that would support a few hundred ip addys in
a
> > > > > guest?
> > > > >
> > > > > i have a few big machines i cant move over until it can efficiently
> > > > > support several hundred ip addys (minimum a /24) ... if ngnet is the
> > > > > answer im gonna try to raise some funding from my boss .. no
promises
> > > > > as he does not let go easily.. :D
> > > >
> > > > hehe, well, ngnet will provide that indeed,
> > > > but for a quick and dirty hack you can raise
> > > > the limit from the current 16 to something
> > > > higher, but you should be aware that the
> > > > current implementation will slow down with
> > > > every ip address you add ...
> > >
> >
> > does anyone have some facts about how slow this would get with 130
> > or 140 ip addys added to a current vserver? is it just measurable or
> > is it visible? unfortunately this mail server is very busy so i am
> > extremely hesitant to try this.
>
> with the current design, the lookup/checks are linear
> so, you can calculate like this:
>
> there is a minimal general overhead, let's say 2
> checks worth on every action
>
> there will be so many checks as are required to
> find a match, so roughly N/2 checks per bind,
> connect, whatever ...
>
> so with N ips assigned, you will get (N/2)+2
> checks, which will give you:
>
> N=1 : 2.5
> N=8 : 6.0 ~ 2.4 times slower
> N=16 : 10.0 ~ 4.0 times slower
> N=256 : 130.0 ~ 52.0 times slower
>
> on those specific checks ...
>
> btw, this will change in the near future, as we
> will switch to something of log(N) order, which
> should eliminate those overheads
>
> HTH,
> Herbert
>
> >
> > ----
> >
> > Chuck
> > _______________________________________________
> > Vserver mailing list
> > Vserver@list.linux-vserver.org
> > http://list.linux-vserver.org/mailman/listinfo/vserver
>

-- 
Chuck
"...and the hordes of M$*ft users descended upon me in their anger,
and asked 'Why do you not get the viruses or the BlueScreensOfDeath
or insecure system troubles and slowness or pay through the nose 
for an OS as *we* do?!!', and I answered...'I use Linux'. "
The Book of John, chapter 1, page 1, and end of book
_______________________________________________
Vserver mailing list
Vserver@list.linux-vserver.org
http://list.linux-vserver.org/mailman/listinfo/vserver
Received on Mon Apr 10 21:45:54 2006
[Next/Previous Months] [Main vserver Project Homepage] [Howto Subscribe/Unsubscribe] [Paul Sladen's vserver stuff]
Generated on Mon 10 Apr 2006 - 21:45:59 BST by hypermail 2.1.8