[AusNOG] Primus Outage explanation

[Curtis Bayne]

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 =)

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).

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.

Food for thought...



-----Original Message-----
From: ausnog-bounces at lists.ausnog.net on behalf of Paul Brooks
Sent: Mon 10/08/2009 12:42
To: ausnog at ausnog.net
Subject: Re: [AusNOG] Primus Outage explanation
 
Without wanting to dampen the flame party, I'll point out that its 
(only!) the people on 240VAC distribution that have problems - the 
carriers that I know were plugged into the -48VDC  supply were 
unaffected this time, and last time also experienced no pain at all.

Yes, the substation has been a bit flaky - but substations do that, 
which is why we assume they will fail and build in backup power 
arrangements for that contingency.
Yes, the gensets have been flaky when they've been called on - but 
gensets are Murphy's favourites, and they do that, no matter how many 
times they are diligently tested.
Its the backup batteries that bridge the gap between mains power failing 
and genset starting that are the real key - and in Primus's case, the 
last failure showed the -48VDC supply has the telco-grade 6 - 8 hours or 
more of reserve capacity, while the 240VAC UPS seems to have only about 
30 minutes of battery. This is the *real* root cause IMHO.

Personally, I try to have at least 4 hours of battery life available - 
which is generally enough time for the gensets to fail to start, get an 
engineer on-site to attempt a manual start, and then kick/swear/curse 
and threaten with escalating sizes of spanners until the genset kicks 
over, or the mains power reappears, whichever comes first.

-- 
Paul Brooks               |         Mob +61 414 366 605
Layer 10 Advisory         |         Ph  +61 2 9402 7355
-------------------------------------------------------
Layer 10 - telecommunications strategy & network design

_______________________________________________
AusNOG mailing list
AusNOG at lists.ausnog.net
http://lists.ausnog.net/mailman/listinfo/ausnog

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ausnog.net/pipermail/ausnog/attachments/20090810/d22604e2/attachment.html>