Hello Grumpa and smeyer and wilbur:
I leave my NASLite+ USB running 24/7. SMART is OFF.
If you turn SMART ON and monitor the "Disk-#.htm" file, it will tell you how many % of self-test remains. I have not run speed tests, and you guys are ahead of me in regards to streaming video ( I currently use only audio (mp3) and archive JPG and other important files ). However, I feel it is extremely likely that at any time while any of your drives have not completed the self-test after power-on you will experience performance issues on intensive (video streaming) operations. For multiple drives, they are running the SMART in parallel (simultaneously), and the time it takes to complete each will depend on the vendor and size and speed of the drive itself.
As far as throughput is concerned: UDMA100, or 66 or 33 or 133 or whatever ... in my experience this number alone is not the sole determining factor. Ofcourse shooting for the higher numbers makes sense ... BUT ... Many other factors of the hardware interact to ding you or give you a bonus in performance. I have direct evidence using UDMA100 motherboards that not only does the CPU and FSB speed make a difference, but much more critically the chipset and the vendors MoBo design ... I have a modern
ECS 800MHz FSB Intel 848P chipset with P4 3.2GHz Prescott and UDMA100 drives that performs at 1/5 the transfer rate when doing a PowerQuest DriveImage Disk-to-Disk (200MB/min) compared to an
Albatron 800FSB Intel865G chipset with P4 2.6GHz Northwood (1,000MB/min). Supposedly the 848 is a poor-mans version of the 865 ... but on my previous system with a genuine
Intel MoBo + P4 1.8GHz Williamette, 400FSB, Intel 845 chipset (old) it ran at 700MB/min ! (Intel Mobo's do well integrating their own chipsets ... as, apparently does Albatron ... not so in the case of ECS, even though they claim UDMA133 (the Intel MoBo is only UDMA100). All drives are UDMA100 spec at 7,200rpm and 8MB buffer size. )
Go figure ... not that I want to muddy the (your) water ... just thought I should share my experience.

Georg