Storage vendors usually rate 7.2k disks for 75, 10k for 120 and 15k for 180. Those are the conservative numbers, in reality with moderate queue depth you see much higher numbers.

They are caused by different viruses, both with several subtypes.
The influenza types usually have much more severe symptoms than parainfluenza.

Speed of light was proven to be a myth. Not interested?

No.
If you replace the daemon binary then wouldn't it make sense to replace a few client binaries as well? :)

http://www.defendec.com/

A little more than what's needed in this case - these things are autonomous for 400 days or 5000 images.

Dropped my S1 about 5 times down concrete stairs bouncing off walls not to mention all the cobblestone and asphalt it has had to deal with. Sure the rim has some minor dents and scratches, the plastic isn't as shiny as it used to be and the screen has a few microscopic scratches (from some nasty type of fine construction sand not any metal objects). Tried banging the screen on an oak table corner a few times when drunk (to show off) and on one occasion dulled a less solid table corner with the same party trick.

I won't try any of that with the S3 but will find out for sure what will it take to break the S1 (aside dropping it to an urinal which fried the usb data, made the microphone act a bit funny at times and rendered the GPS aerial quite useless on a city street).

Would love to see a bit more scientific drop test video of the S3 than the one everyone's referring to. Just to feel safe (or not).

regarding power supplies .. blade servers have one set of power supplies per enclosure or even rack - maintaining redundancy is much more efficient that way. In addition to ouputting DC I bet all of them are available with DC input also. In a facility with central DC supply the total savings are certainly noteworthy. Networking and cooling are also shared and more robust, probably making the whole lot (energy and materials wise) cheaper to manufacture and run. Overall, blades inherently tend to use more energy-efficient components because of the increased density.

That said I'm not sure that a significant percentage of real world deployments can make use of all this. Plus depending on the IPC needs of the split up applications the overhead of communication and stalls between blades may outweigh all of the above. Especially since we don't have general purpose asynchronous chips yet - which reminds me .. where are the chips promised by Sun several years ago? :)

But "microblades" .. as already speculated the whole blade concept may steer towards a robust and easily scalable, massively-SMP design. The spiral continues ...