The dot in ... --- ...

Chris Gerhard's Weblog

« Previous day (Mar 7, 2005) | Main | Next day (Mar 8, 2005) »

20050308 Tuesday March 08, 2005

Bonked again whilst cycling home last night.

It happened again last night. Got about 8 miles home and had to consume the emergency energy drink which got me almost all the way home before the energy just ran out. I'm beginning to get a bit concerned that I'm not getting the right kind of fuel at the right time despite my best efforts.


( Mar 08 2005, 05:04:46 PM GMT ) Permalink Trackback

   
Disk Scrubbing

I wrote and maintain a disk test tool that gets used for all sorts of purposes for which it was never intended. This week the question of disk scrubbing came up and I was surprised to see just how fast it could scrub a minnow. Disk scrubbing is a uniquely strange form of IO as it is all about writing as much data as possible and you never need to read the data back. But being able to scrub at a rate of 170Mbytes/sec to a single device via a single connection was pleasantly surprising. I guess I just have not been looking that hard before. Any way it gave me a n excuse to try the USCSI option to the disk tester in anger so it was worth writing after all!

What is odd is the misconceptions about what is needed and when from a disk scrubber. If you are using a disk in a secure data centre then there is no need to scrub the disk with multiple passes, which prevents at least hinders techniques like magnetic force microscopy, since those techniques require physical access to the disk drive. You have already accepted that the data centre is physically secure otherwise your data should never have gone there.

The issue is now just a question of how your data can be protected from over the network access. In that case if the disks are going to be reused by another user they would need to be scrubbed, just once, to prevent the data being read remotely by a user who now has read write access to the disk but does not have physical access. If the disk was to accept a firmware down load, then it might be possible to still do harm, but if the disk in question is in fact an array that can not be configured from the host computer then just a single write will do.




( Mar 08 2005, 04:58:13 PM GMT ) Permalink
Trackback

   

Valid HTML! Valid CSS!

Except where otherwise noted, this site is
licensed under a Creative Commons License 2.0

This is a personal weblog, I do not speak for my employer.