[36351] in North American Network Operators' Group
RE: Faster 'Net growth rate raises fears about routers
daemon@ATHENA.MIT.EDU (Roeland Meyer)
Tue Apr 3 15:42:47 2001
Message-ID: <9DC8BBAD4FF100408FC7D18D1F092286039DC8@condor.mhsc.com>
From: Roeland Meyer <rmeyer@mhsc.com>
To: "'Eric A. Hall'" <ehall@ehsco.com>
Cc: nanog@merit.edu
Date: Tue, 3 Apr 2001 12:36:27 -0700
MIME-Version: 1.0
Content-Type: text/plain;
charset="iso-8859-1"
Errors-To: owner-nanog-outgoing@merit.edu
> From: Eric A. Hall [mailto:ehall@ehsco.com]
> Sent: Tuesday, April 03, 2001 12:10 PM
> Convincing customers that it is cheaper/better to put their
> main servers
> somewhere off-site away from them is the challenge. Otherwise
> more of them
> would do it.
You understate the problem. Service bureaus and ASPs have one problem in
common, MIS/IT departments are loath to place their corporate data (jewels)
under the control of another. Convincing them otherwise is an up-hill battle
that you will mostly lose.
To a lesser extent, this is also true of colos. The other part of the
argument, that the local (fully burdened) square-footage is almost always
cheaper than decent colo space ($1,200.00 per 1/4 rack, per month, typical).
You therefore have two strong motives for locating corp datacenters on-site,
rather than off-site. Either one, by themselves, are difficult to overcome.
Then you have the third argument; most corporate data centers are used more
internally than externally (road-warriors and B2B ops not withstanding). For
performance reasons, you want the servers closest to their users. Unless you
can get 100baseTX to the off-site facility, this means a strong incentive to
mount servers at HQ, not in the colo.
We are now back to multi-homing a business. BTW, this entire mess was played
out before, in the 70's and 80's ... when the service bureau biz died. At
that time SNA APPN/APPC was the network (my briefs were dyed "blue" at the
time). The end-game was proprietary pre-internet data centers, connected by
mutiple slow links.
