home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!wupost!uwm.edu!psuvax1!psuvm!auvm!USCMVSA.BITNET!LDW
- Message-ID: <IBM-MAIN%92090803032623@RICEVM1.RICE.EDU>
- Newsgroups: bit.listserv.ibm-main
- Date: Tue, 8 Sep 1992 01:00:00 PDT
- Sender: IBM Mainframe Discussion list <IBM-MAIN@RICEVM1.BITNET>
- From: Leonard D Woren <LDW@USCMVSA.BITNET>
- Subject: Re: PROC TAPELABEL and 3490E drives
- Lines: 83
-
- > The original 3480's were:
- > 1.) 18 track
- > 2.) 550 feet long
-
- We've been over this before on this list...
-
- The manual states that the usable area of a 3480 cartridge must be at
- least 150 meters. That's 492 feet. The fact that some vendors are
- selling longer cartridges is an aside. A "normal" length cartridge is
- typically 3-5% longer than the 492 feet, making it 510 or so feet.
- All of the cartridges in *our* library that I checked when I was
- adding the IDRC code to TAPEMAP showed 103-105% full, with TAPEMAP
- using 492 feet as 100%. So I have faith in my (and TAPEMAP's)
- arithmetic.
-
- > 3.) 38k bpi (density)
-
- Actually, according to the manual, it's 1491 bytes/mm, which is 37871
- bytes/inch.
-
- > 4.) .08 inch inter-block gap
-
- I don't think the IBG is given in the manual, but everyone seems to
- agree that it's .08 inch. What I'd still like to know is how much
- tape a tapemark takes. The reason why I'm concerned about this is
- that on a 9 track, a tapemark is followed by a 3.75 inch erase gap.
-
- > and a CART could hold about 240 mb at max.
-
- The manual states 200 mb. This is no doubt figured using the minimum
- required length, and a lower than maximum blksize.
-
- > I'm told that a later incarnation of the 3480 implemented data compression
- > with hardware. Was it different in any other way?
-
- I don't know if IDRC (Improved Data Recording (what does the C stand
- for?)) is a new version so much as a later add-on feature for 3480s.
-
- > What were the major differences of the original 3490 drives?
-
- IBM fixed the design problems in the 3480? :-)
- IDRC is standard, I think.
-
- > Did the Blue gentleman say that the 3490E CART is 1100 feet long and that
- > the drive reads/writes 18 tracks in each tape direction? With double the
-
- Well, the 3490E _does_ write 18 tracks in each direction. This means
- that if you're filling a series of cartridges, the rewind time is
- zero! I had heard somewhere that there would be double length
- cartridges available. Can someone with 3490Es confirm this?
-
- > physical length and double the number of tracks (total) of the old 3480,
- > does this mean that a 3490E CART can hold around 960 mb BEFORE compression?
-
- Probably more like 800mb or so, depending on what length you figure on
- and the blksize. Note that a blksize of 3040 on a cartridge wastes
- HALF of the tape in IBGs. IEHMOVE's hardcoded output blksize of 800
- wastes 80% of the cartridge. A blksize of 80 (unblocked card images)
- wastes 99.8% of the cartridge. 32K blksize is normally appropriate
- (and only wastes 10%...) FDR (and maybe DFDSS; I don't remember) uses
- larger block sizes, in order to fit a whole track image into one tape
- block.
-
- Now, I have a question. Does IDRC use very large blocks (to reduce
- the amount of tape wasted in IBGs?)
-
- > If so, I, for one, am impressed.
-
- So am I. Now, if you figure an "average" compression ration of 3.5:1,
- one of those suckers should just about hold a full dump of a 3390-3.
- IBM was getting concerned about how many volumes were required to dump
- their largest disks. It looks like the 3490E addresses that problem.
-
-
- For those familiar with GIF, an interesting data point. GIFs are
- supposedly compressed using a fairly good algorithm. And I'd guess
- that IDRC uses a fairly good algorithm. I put a bunch of GIFs on a
- cartridge for a friend of mine, and just for amusement I specified
- compression. My new TAPEMAP said "overall compression factor for
- this tape is 1.0"! I've seen other files with compression factors up
- to 11, so this was somewhat amusing.
-
- /Leonard
-