I'm at work. PG&E decided we had too clean of power for too long and gave us a power hit at 7:20 this morning. We don't have UPSes for the main systems, so I've been here repairing.
Why? You cry! That's a pretty simple thing to explain. Except for a few critical things that require us to have up and running except for when the building falls down, we checked the cost of how much time it would buy us (on average) per year of down time vs the increased number of CPU cycles. The CPU cycles saved from the down time was far less than the CPU cycles that could be purchased by buying computer hardware instead of UPSes.
Now that makes sense. It blatantly sucks for those on rotation when we get a power hit though. Like, oh, me!
Right now the IBM has been healed (mostly). A Linux cluster is being healed albeit partially because lustre sux raw eggs. (or at least the version they're using).
I've been here since about 800 AM Pacific. Sux, I tell ya. This is not POC rotation. This is Hell Week. I am going to see if I can get this renamed at the turnover meeting on Tuesday.
Why? You cry! That's a pretty simple thing to explain. Except for a few critical things that require us to have up and running except for when the building falls down, we checked the cost of how much time it would buy us (on average) per year of down time vs the increased number of CPU cycles. The CPU cycles saved from the down time was far less than the CPU cycles that could be purchased by buying computer hardware instead of UPSes.
Now that makes sense. It blatantly sucks for those on rotation when we get a power hit though. Like, oh, me!
Right now the IBM has been healed (mostly). A Linux cluster is being healed albeit partially because lustre sux raw eggs. (or at least the version they're using).
I've been here since about 800 AM Pacific. Sux, I tell ya. This is not POC rotation. This is Hell Week. I am going to see if I can get this renamed at the turnover meeting on Tuesday.
No comments:
Post a Comment