[62948] in North American Network Operators' Group
Re: "Class A Data Center"
daemon@ATHENA.MIT.EDU (Scott Francis)
Wed Sep 24 14:39:57 2003
Date: Wed, 24 Sep 2003 11:20:25 -0700
From: Scott Francis <darkuncle@darkuncle.net>
To: Bob German <bobgerman@irides.com>
Cc: "'Jay Hennigan'" <jay@west.net>, nanog@merit.edu
Mail-Followup-To: Bob German <bobgerman@irides.com>,
'Jay Hennigan' <jay@west.net>, nanog@merit.edu
In-Reply-To: <009d01c37e1f$3c3c8130$6500a8c0@LittleKahuna>
Errors-To: owner-nanog-outgoing@merit.edu
--sdtB3X0nJg68CQEu
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Thu, Sep 18, 2003 at 03:58:31PM -0400, bobgerman@irides.com said:
>=20
>=20
> This is the assumption I have come to as well. Are there any
> established standards for enterprise datacenters at all, aside from the
> obvious, N+1 redundant everything, diverse paths, etc.?
I don't know if it qualifies as an "established standard", but ISTR that
Steve Bellovin had a paper about various levels of reliability in data
centers ... [searches] argh. I can't find it yet. Perhaps Mr. Bellovin can
refresh my memory ... the paper I'm recalling had specifications for 5 or so
different levels of reliability and redundancy in data centers (able to
withstand criminal attack, armed attack, conventional explosives, nuclear
explosion, acts of God, etc.) and was interesting reading. The focus, as I
recall, was on the level of engineering required to reach various levels of
uptime (99.9, 99.99, 99.999, etc.).
This ringing a bell for anyone else?
--=20
Scott Francis || darkuncle (at) darkuncle (dot) net
illum oportet crescere me autem minui
--sdtB3X0nJg68CQEu
Content-Type: application/pgp-signature
Content-Disposition: inline
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.1 (FreeBSD)
iD8DBQE/ceBpWaB7jFU39ScRAmE3AKCAcR+dx4RKO/kb68tzOiQdNAvN6ACeJING
yNMsGD9SMuixn9j+TObBUww=
=Ffdl
-----END PGP SIGNATURE-----
--sdtB3X0nJg68CQEu--
