[mod.computers.vax] Forwarded message for Tapes

ANK@CUNYVMS1.BITNET.UUCP (03/04/87)

From:   UK%"NETWORK%UK.AC.ULSTER.UUVAX@AC.UK"  3-MAR-1987 06:55
To:     ANK
Subj:   files expand when copied to disk from tape

Received: From UKACRL(MAILER) by CUNYVMS1 with RSCS id 5629
          for ANK@CUNYVMS1; Tue,  3-MAR-1987 06:55 EST
Received: By UK.AC.RL.IB (MAILER) ; Tue, 03 Mar 87 11:48:25 GMT
Via:      UK.AC.ULST.UUVAX;  3 MAR 87 11:48:20 GMT
Date:     3-MAR-1987 11:35:30
From:     NETWORK%UK.AC.ULSTER.UUVAX@AC.UK
To:       ANK@CUNYVMS1
Subject:  files expand when copied to disk from tape

No they don't. Blocks on tape are not always 512 bytes, as they are
on VAX disk. In fact you get 1 tape block per "cluster" where the
cluster size is taken from the disk volume from which the file was
copied to tape (assuming the tape was created with VMS COPY). If
the cluster size on the (source) disk was 8 then a directory of a
tape file will have 1 block for each 8 on the disk, plus 1 extra
for any oddments left over in the last cluster. Copying back will
restore the file exactly as it was (may be *marginally* smaller if
the cluster size on the target disk is different, only noticeable
if you do DIR/SIZ=ALL - the difference is simply in the number of
blocks required to make up the last cluster).

If you go round deliberately using large block sizes on tapes (we
commonly use 32000 bytes or 64 disk blocks) the apparent discrepancy
is WORSE. However this gives faster tape I/O and uses less tape since
you read/write far fewer inter-block gaps.

Hope this explains things a bit. Please pass this on to the net, I'm
readonly to arpa.
                                                Brian Beesley
                                                SCO Data Comms
                                                Univ. of Ulster
BITNET: BRIAN @ UUVAX.ULSTER.AC.UK              Northern Ireland
(via UKACRL)