ral@cathedral.cerc.wvu.wvnet.edu (Rebecca A. Littleton) (11/26/90)
We are using the BRU command for backups on our SGI file server onto a Kennedy Model 9610/9660 tape drive (1/2" magnetic reel-to-reel drive). Backups are taking 1 hour to fill each tape (150 M tape). This is too slooow and not practical. If you are using BRU for backups onto reel-to-reel tapes, does it take you this long? If it doesn't, do you know what I might change or what I might be doing wrong? I have thought of: (1) bad /etc/brutab entry. I use the default supplied by SGI. /dev/rmt/xmt0d0.6250 | /dev/mt/xmt0d0.6250 \ size=0K seek=0 \ prerr=0 pwerr=0 zrerr=0 zwerr=0 frerr=0 fwerr=0 wperr=0 \ reopen rawtape tape advance /dev/rmt/xmt0d0nr.6250 | /dev/mt/xmt0d0nr.6250 \ size=0K seek=0 \ prerr=0 pwerr=0 zrerr=0 zwerr=0 frerr=0 fwerr=0 wperr=0 \ norewind reopen rawtape tape advance (2) bad tape drive. It checked OK on setup by SGI. (3) BRU is inherently slow. (4) porting the dump command (a la SunOS dump) to the SGI. If you have increased the speed of BRU, how did you do it? I would really appreciate knowing how I can speed up backups. We have 5G of disk space on our SGI server and are currently using 14 tapes for a monthly backup. HELP!!!
mg@ (Mike Gigante) (11/27/90)
I don't know about 1/2" tapes, but on our exabyte, we sped bru up *considerably* by increasing the buffer size to ~ 1Mb. Unfortunately I have found the double-buffering option to bru to be unreliable as this would be even faster. e.g. bru -cv -b 1Mb -f /dev/xtape ... the larger buffer reduces all the back and forth tape motion that is needed to get to streaming speed (for a little 20k write). buffers larger than 1Mb didn't seem to win much. (This was a very qualitative trial!!) Mike Gigante, RMIT Australia
karron@KARRON.MED.NYU.EDU ("Dan Karron ") (11/27/90)
Check out bstream to buffer stuff to your device so it can stream. I have found that some 9tracks (Cipher) reel back and forth to write data that is not blocked properly. With bstream blocking, the tape only moves forward. I noted previously here that there were NO sgi entries in the /etc/brutab. That seems silly. If sgi is going to support bru, they would at least let bru know about the devices it had brewing. Someone should clue bru into doing the write thing with a new bru tab. (I could not resist rhyming) Cheers! dan. +-----------------------------------------------------------------------------+ | karron@nyu.edu (E-mail alias that will always find me) | | * Dan Karron, Research Associate | | . . . . . . . . . . . . . . * New York University Medical Center | | 560 First Avenue \*\ Pager <1> (212) 397 9330 | | New York, New York 10016 \**\ <2> 10896 <3> <your-number-here> | | (212) 340 5210 \***\_________________________________________ | | Main machine: karron.med.nyu.edu (128.122.135.3) IRIS 85GT | +-----------------------------------------------------------------------------+