I had thought that flash devices had higher throughput than HDDs, but I see that I was wrong. They do have faster seek times (this advantage is greatly mitigated by the presence of cache on a HDD), but the throughput is somewhat less. I’m pretty sure that the fastest flash devices are faster than USB 2.0, so the difference only matters when considering flash as a replacement for normal hard drives.
In the cases where flash is likely to replace hard drives, power consumption tends to be king, even to the point where hard drives are slower models, and are spun down when possible. In that kind of application, flash may well have higher average throughput for relatively large files.
It may also be more economical to parallelize flash than HDDs, too. Generally, multi-platter drives cannot read/write to multiple platters at once, but even if they could, you run into physical limits with respect to the number of platters a disk can have. A 3.5" flash device could probably have an order of magnitude or two more parallel devices in it, and the throughput should increase linearly.
Microsoft are pushing Hybrid drives for Vista - with the Flash on the drive being used for hibernation/rapid restart. Manufacturers will probably continue to increase the flash storage on these drives, and bits of the OS may move to the flash storage as well.
Interestingly, Vista also lets you allocate space on any flash device for OS use, so shoving a 4Gb SD card in a slot to be used as a semi-permanent system cache sounds like a good way to boost a system struggling with MS bloatware :smack: software.
My Big Idea is an encrypting file system for roadwarrior laptops that stores a semipermanent journal on internal flash. The journal gets reset via a backup process (which only needs to transfer the journal data to the backup server). If the drive fails, the journal can be used to recreate the latest data.
A question based on this. Is this 10,000 per location or over all? Or maybe it’s better to ask the question this way:
Suppose I have 4 Gb flash drive, but on average only have a few Mb of stuff on it. If I’m constantly changing the stuff (e.g., use it daily to take new files from home to office and vice versa) is it better to not erase the older material, because then the new material will constantly be written in the same physical spots leading to earlier failure? Should I let the flash slowly fill up and then erase the entire thing and start over?