Page is a not externally linkable
piatkow - 10:17 am on Jul 8, 2009 (gmt 0)
An engineer tried to free a cable from the underfloor spagetti - guess which two power leads were disconnected? Recovery plans are never seem to be tested properly and only ever seem to cover a sub set of possible issues. Of course with 24*7 systems literally pulling the plug on the live system on a Saturday afternoon to see what happens is no longer an option.
At my last company (about 12 years ago) our NonStop Tandem system was supposed to do that. As long as two specific servers weren't taken down at the same moment.