Actually, I argued that the DV tape did have not only the capacity but might make plenty of sense for digital still use.... not the other way around.
However, on closer examination it looks like in the space you can get tape, you can get a hell of a lot more disk!
The funny thing is that the same thing happened with DLT in the data world. DLT, prior to the "super" varient that is way new and with an unproven history, used to hold a LOT more on a cartridge than a disk of approximately equal physical size. Well, not any more! Now that same cartridge holds 70GB (compressed, @ 2:1) but I can buy a disk quite cheaply (for about 4x the cost of the tape!) that holds 160GB, and 320GB disks are available in the same formfactor, albiet they are expensive. Even the "Super DLT" though has been surpassed with the disk drive capacities in the same physical size!
It no longer makes economic sense to use tape to back up disk due to transfer rate and storage density limitations. Rather, it now makes sense to use RAID arrays for physical redundancy and to back up to a SECOND array, which then has the "capsules" (drives) pulled and stored off-site in a vault. Its both faster AND cheaper to do this than to buy the tape library and media to do it to cartridges with a proper tape rotation! Optical or opto-magnetic media, which used to be all the rage due to random access capabilities, is now a bad (and expensive) joke.
The last database storage array I built for a customer (I still do consulting work in this realm), totalling close to a terabyte of online storage, was a Raid-5 array backed up with a second one in exactly this fashion. The second one was then pulled and placed in a vault offsite, and the carriers replaced with the other copy. This rotation was done every week, so in the event of a fire or other catastrophic event, no more than one week's worth of data would be lost and have to be recreated. That was an acceptable risk for that customer. It was VASTLY cheaper to do this than to do the same job to DLT library when hardware, software and media costs were considered. The full copy required about 9 hours (including the snapshot functionality and roll-forward when the backup was complete), and was run overnight Sunday, assuming the source array was full. At "usual" load, which is about half full, it was a 4-hour job. The beauty of this system is that in the event of a fire the backup media is instantly usable and requires no reloading back to working storage, as it IS working storage, its about 4x faster in performance than the same backup done to SuperDLT, and the database doesn't have to go offline - ever!
I pulled that 8gb drive I have. It won't QUITE fit in a CF-II form factor, but its pretty close. Then realize that the drive is from the late 90s, and things become more compelling.
BTW, the Hitachi drive you referenced is cited as having MORE speed than any CF card on the market. If true, then it will handily support a DV video rate. I believe the claims, as most of the gains in I/O speed on Winchesters have not come from the increased platter rotational rate, but rather from the increased areal density of the bits on the platter.
Nor is this about to end. IBM has had in development for several years now technology that will record the magnetic domains "vertically" rather than "horizontally" on the platter, and they are also experimenting with phase-angle encoding. The gains in areal density that can be had with these techniques are almost beyond comprehension If they get it working in production multi-terabyte capacity 3.5" formfactor drives will be quite realistic, and 100GB+ 1" drives with transfer rates rivalling the best of what we have now in the 3.5" form factor will be not only possible, but commonplace. The idea of a wristwatch or PDA - literally - with more storage in it than you currently can fit in your desktop machine is NOT an impossibility.
The Ridata 1GB card that I have, which some say is the "fastest" card out there, can handily handle data rates of over 5MB/sec in read mode, and about 4MB/sec in write. It is BLAZING fast, and will definitely handle a direct capture from a DVCam; I've done it as a test (until it ran out of space!) and had no dropped frames.
BTW I took a 2GB AVI file this evening (a DV capture of about 16 minutes) and compressed it with gzip, to see how "routine, ordinary" compression tools do on such data. This is a lossless compression that has no knowledge of the contents of the file, and thus can't "cheat" in any material way. The source material was a band halftime show clip in 16x9 format.
I timed the compression on my Unix box, which is a 1.6ghz AMD chip; the job was saturated on the disk channel (on a 160GB WD disk - a cheapie, one of my $139 specials recently) at 13 MB/sec (!) - so much for those "cheapie disks" being slow. This is a FILESYSTEM I/O run; just for comparison Windows can't TOUCH that data rate, not even XP on a 2.4Ghz machine (so much for XP's "prowess" with NTFS, etc?)
I was consuming about 80% of the CPU on this job, so I still had some CPU remaining. The job required approximately 5.5 minutes, but led only to a savings of about 20% of the storage - far less than I had expected. Whether this is due to the "AVI" format I do not know, but it is significantly less gain than I had expected.
For grins and giggles I then ran a disk performance benchmark test that I have laying around (which I wrote several years ago for use at a client site) that measures performance across the entire disk's surface as a raw I/O data rate. I haven't used this on a disk in years - previously, 5-6MB/sec was a pretty good performance turned in on SCSI drives that were being tested at the time.
That SAME disk records a RAW I/O speed of 42(!) MB/sec (that's bytes, not bits!) when timed with "dd" (a direct I/O to and from the platter.)
42MB/sec is, quite honestly, mind-blowing. It is indeed, however, what kind of sequential SUSTAINED performance I can get from that disk. That's roughly half of the ATA/100's RAW possible bus data rate - damn impressive. To put this in perspective that would sustain TEN DV streams AT ONCE and not drop a single frame! From a 160GB disk that cost me $139 at Sam's Club - I'm flatly astounded.
)
) Out of those, essentially all were gravitating towards the two camcorders that recorded directly onto the little DVD recordable disks. Why? Convenience. It won over performance. Yeeeccchhhh says I, but Granny wants convenience for shooting movies of her kids school play!
