Sorta. They had n+1 They took down the +1 for regular maintenance leaving them n. Then one failed leaving them n-1. Then the remaining unit could not carry the load and shutdown to protect itself. I'm not about to preach to Transpower, anyone who saw Richard Naylors talk at NZNOG will appreciate the huge job they have. But the lessons for us network people must be: If the chance of one link/router failing is x. Then if you have 3 parallel/resilient links/routers then the risk of a total outage is x/3. If you take one of those links/routers out for maintenance then the risk of a total outage is x/2. But here is the kicker. If the single remaining link/router can't take the total load without causing such significant loss or interruption that it had may as well be down... then the risk of a total outage with 2 links is really x. Have a look around your network. The same thing applies to UPSs and all sorts of things. Feel free to flame if I have my maths/stats/facts wrong. Dean Matthew Poole wrote:
Nope, n+1, then the n took down the +1. Otherwise it wouldn't be able to operate on a single transformer. So we'd better hope that, until the +1 is fixed, n behaves itself. My guess would be that it's roughly n+2 at this time of year, but only n+1 during winter, or maybe even once the universities all get back up to full speed in three more weeks.
_______________________________________________ NZNOG mailing list NZNOG(a)list.waikato.ac.nz http://list.waikato.ac.nz/mailman/listinfo/nznog