From: Dean Strik (deanstack.nl)
Date: Mon Feb 02 2004 - 10:25:25 CST

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Victor.Duchovnimorganstanley.com wrote:
> On Mon, 2 Feb 2004, Dean Strik wrote:
>
> > If 40%-60% is unacceptable (I find that hard to believe), then yes,
> > buying expensive load-balancing equipment to make it nearer 50-50 is an
> > option. So is building some simple load balancing box itself. But
> > personally, if I had to spend money on new hardware for load balancing,
> > I'd just buy a third machine - distributing load over 3 boxes..
> >
>
> I am seeing a volume independent 52%/48% split in favour of the first of
> two MX records. The sample size is ~500000 messages, so the standard
> deviation for a fair binomial distribution is ~0.07%, this means that ~2%
> of the clients that connect to Morgan Stanley's Internet facing MX hosts
> don't shuffle the MX records. I don't know whether this assymetry rate is
> representative of other high volume sites, or just peculiar to us.

This sounds about right. I see about the same for our university network
(not counting mail from internal sources): 53%/47%.

--
Dean C. Strik Eindhoven University of Technology
deanstack.nl | deanipnet6.org | http://www.ipnet6.org/
"This isn't right. This isn't even wrong." -- Wolfgang Pauli

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]