<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 3.2//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<META NAME="Generator" CONTENT="MS Exchange Server version 6.5.7638.1">
<TITLE>RE: [AusNOG] Primus Outage explanation</TITLE>
</HEAD>
<BODY>
<!-- Converted from text/plain format -->
<P><FONT SIZE=2>By which time your servers are usually cooked because you're on the 3rd story of a building and unable to open any doors/windows to get airflow =)<BR>
<BR>
The reality is, a lot of the green datacentre initiatives actually IMPROVE availability (side air economisation, hot aisle containment) by minimizing the impact of a failed mechanical system. I struggle to understand why smaller facilities aren't implimenting them (and saving themselves a bunch of money on electricity to boot).<BR>
<BR>
As I mentioned in the Whirlpool thread, for datacentres of this size (1000KVA and below) it nearly makes sense to pull an LV drop rather than having a dedicated demarcation substation - whilst this lacks the wank factor of having your own transformer (look ma, no competition!) it does have the added advantage of distributing your risk between the 15-20 odd 350-1000KVA step down transformers that feed your local grid.<BR>
<BR>
Food for thought...<BR>
<BR>
<BR>
<BR>
-----Original Message-----<BR>
From: ausnog-bounces@lists.ausnog.net on behalf of Paul Brooks<BR>
Sent: Mon 10/08/2009 12:42<BR>
To: ausnog@ausnog.net<BR>
Subject: Re: [AusNOG] Primus Outage explanation<BR>
<BR>
Without wanting to dampen the flame party, I'll point out that its<BR>
(only!) the people on 240VAC distribution that have problems - the<BR>
carriers that I know were plugged into the -48VDC supply were<BR>
unaffected this time, and last time also experienced no pain at all.<BR>
<BR>
Yes, the substation has been a bit flaky - but substations do that,<BR>
which is why we assume they will fail and build in backup power<BR>
arrangements for that contingency.<BR>
Yes, the gensets have been flaky when they've been called on - but<BR>
gensets are Murphy's favourites, and they do that, no matter how many<BR>
times they are diligently tested.<BR>
Its the backup batteries that bridge the gap between mains power failing<BR>
and genset starting that are the real key - and in Primus's case, the<BR>
last failure showed the -48VDC supply has the telco-grade 6 - 8 hours or<BR>
more of reserve capacity, while the 240VAC UPS seems to have only about<BR>
30 minutes of battery. This is the *real* root cause IMHO.<BR>
<BR>
Personally, I try to have at least 4 hours of battery life available -<BR>
which is generally enough time for the gensets to fail to start, get an<BR>
engineer on-site to attempt a manual start, and then kick/swear/curse<BR>
and threaten with escalating sizes of spanners until the genset kicks<BR>
over, or the mains power reappears, whichever comes first.<BR>
<BR>
--<BR>
Paul Brooks | Mob +61 414 366 605<BR>
Layer 10 Advisory | Ph +61 2 9402 7355<BR>
-------------------------------------------------------<BR>
Layer 10 - telecommunications strategy & network design<BR>
<BR>
_______________________________________________<BR>
AusNOG mailing list<BR>
AusNOG@lists.ausnog.net<BR>
<A HREF="http://lists.ausnog.net/mailman/listinfo/ausnog">http://lists.ausnog.net/mailman/listinfo/ausnog</A><BR>
<BR>
</FONT>
</P>
</BODY>
</HTML>