My big plan to "live in the cache" and automatically consolidate my offline changes when I got my tablet back backfired. If you remember, I have my working notebook stored on my Gateway tablet, and share the folder it is in over the domain here at work. I have it open from 2 other machines since I roam between different locations throughout the course of a day.
When the Gateway died, I lost the "real" copy of the notebook. When it came back from the shop recently, I was hoping the hard drive had not been wiped. It hadn't - as near as I can tell, the screen had gone out and been replaced. I powered it up but it could not be seen by any other machines on the network. Sigh - I got to start troubleshooting basic network connectivity instead of completing my testing. To make a very long story short, somehow the IP address was set to be static, and that was causing all kinds of errors. By the time I had that fixed, I had manually copied my changes over via sneakernet. It's frustrating to lose out on testing due to problems like this - I can only suppose Gateway had to change some setting when they had it to complete their testing.
Since then, I decided to move my notebook from the local hard drive on the tabley and have begun putting it on an SD card. So far, so good, and now I can carry it with me wherever I go. Not quite as universal as a USB stick, but the SD card doubles as a ReadyBoost cache and I get a slight speed improvement. Plus, it doesn't stick out as far as a USB device and is much easier to carry around while plugged in.
This does bring up an interesting (to test) tangent about flash devices.
The most interesting test with this scenario is performance. You would think that avoiding a hard drive and having a flash drive (essentially what I am using) would be significantly faster than a spinning hard disk. This is not quite the case. Flash devices are much slower to write to: they have to complete a full read before they can write back, so the write cycle takes longer. Reading data from the device is much quicker, and overall this results in a slight speed boost. A task for the test team here in Office will be to quantify our performance with flash devices and flash drives. The new SSDs have drivers and interfaces that do all kinds of work under the hood - minimize writes, spread the data around to prevent burnout, etc… and there is a wide variance in the SSDs - some have write times of over 200 ms (technically, a "non-contiguous write time prep value"). This is roughly equivalent to a 4200 RPM hard drive - not a speed demon by 2008 standards. Some are much faster and you can think of them as "5400 rpm" equivalent drives. Since they are starting to be offered in more models of laptops, we will start quantifying our performance on them.
Questions, comments, concerns and criticism always welcome,