Not about Makerbot or Skeinforge stuff..
So, full disclosure, I work at Stratus Technologies. We make high availability and continuous availability servers – servers that are about as rock solid and reliable as you can get. We’ve got software that crawls our support database every day – the machines I write code for (ftServer) are at 99.99971% availability as of today, which is damn good.
Hey! I may be speaking from Stratus but I’m not speaking for Stratus. This is “day in the life of..” not “press release..”
So, they’re pretty cool machines. Fully redundant hardware means no single-points-of-failure on the machine. Running processors in lockstep means you really don’t get hit with any failover time if something goes wrong. None of that torture you get with installing & managing clusters. Single OS license. Good stuff.
The other side of the isle my office is in is populated by hardware folks and they’re all excited about a video they made, using some proto hardware that was headed for the dumpster – stuff we got really early in the release schedule (so we can make it work) that’s really not production quality and shouldn’t be sent to customers.
They should have gotten a marketing person to do the video. It’s a bit shaky and, well, they’re very much not actors – they’re nerds. Sorry, guys but it’s true. It’s true for us software people too, so don’t feel bad.
Anyway, it’s an amusing video. It starts with them showing some generic, non-Stratus server with the covers off. There’s some heartbeat type graphics running on the monitor it’s attached to. One of the hardware guys takes a cup full of nuts & bolts and dumps it right down on the motherboard. Then he takes a cup of coffee and dumps that in there, too.
Not surprisingly, you can see the screen say “signal lost” as soon as the nuts & bolts hit the board. Sadly, there’s no smoke and nothing catches fire, which would have been a nice touch. That machine is toast though, and it’s never going to get better.
Next up, they repeat the test on one of our (two 4U rack modules, connected by a backplane) systems. One module has the covers off and they’re showing the same heartbeat-type application along with our monitoring software which shows the state of the machine.
The cups of nuts, bolts and coffee go down on the motherboard but the heartbeat video keeps on beating. You can see from the monitoring software that something’s unhappy but the machine’s built to detect unhappy components and take them offline – the application never even notices.
The hardware guys say they’re trying to get the video onto youtube. I’ll post an link if I see it appear..
I also heard mention of a “high powered rifle” and the rod & gun club just down the street – apparently they have some more hardware that’s not needed any more. I’m torn between asking if I can watch and wanting to stay here, safe, in the middle of a big, solid building. I have no doubt the machine would survive, though…