When someone announces a downtime to me and puts a time limit on it (7am to 12 noon PST, to use popular *snicker* example), I expect to see an available service soon after the announced downtime window finishes, not routine just-under-the-wire announcements as to what problem or other there is with bringing it back up followed by huge expansions of the original downtime window. Once or twice is workable, beyond that it starts to suggest serious systemic problems with estimating and completing work to a fixed schedule. I suppose not having to enforce or adhere to any kind of SLA helps.
Really though, you need to get better at this. You aren't getting better, which is why you routinely overrun your downtime windows by levels that would be unacceptable in almost any sensible technology business model. So, LL, what are you doing to fix this situation? It WILL catch up with you some day, and I'd hate to be you when that happens.