They should start...
A "worlds crappiest microphone" competition...
Because everything seems to have ears these days...
It must be one of the worst ways to build a microphone imaginable, tapping into a disk drive's nanosecond head stops as it waits for the vibrations caused by sound to cease, but it has been done. When sound meets a disk drive, the sonic vibrations cause a shaking in the disk drive and the platters vibrate. The drive's control …
This is very interesting. This means that if I apply noise cancellation technology used in headphones I can prevent the hard drive from stopping getting higher throughput from existing hard drives. Wonder if this can be done by an external device or if I need to work with the hard drive manufacturer to get it built in.
Active noise cancellation might be useless, but isolating noisy components in the datacenter and soundproofing everything else might help.
Of coursesound proofing necessarily blocks or restricts ventilation, so maybe a combination of soundproofing with liquid cooling. Is that too complex/expensive to improve disk performance. And how much improvement are we talking here? article doesn't specify but I would guess that if noise were a major problem interfering with disks, I would have heard a lot of noise about it :)
Well I suppose if you wanted to shut down a server room without any visible sign of damage to the servers (assuming they still run HDD) this might have its uses.
Why would you want to do that other than some elaborate BOFH plan?
I probably still have the IBM 1130 assembler code that sent signals to the big old Winchester “washing machine” disk drives at different frequencies. The program took input in the form “AABBC+” etc. to form musical output. Output was generated by setting a transistor radio on the console above where the channel wires were routed. The signals in the wires were powerful enough to generate sounds on an A.M. radio nearby.
Needless to say this was probably not good for the Winchester drive!
A device you can't turn off, that may (or may not) be listening to you at any minute of the day.
Except companies have persuaded people that this is necessary and that they want it.
IRL watches and phones have been able to do short real time speech recognition (IE the "wake up" bit) for decades.
Today I fired up some old disk shelves that made me adjust the labs noise-rating. Previously we were at "bloody loud", but after accounting for the extra equipment we now have a new designation of "DID YOU SAY SOMETHING?".
Are you telling me the slow performance is due to the noise? I thought it was because of the 10 year old disks...
Biting the hand that feeds IT © 1998–2019