Article View: comp.periphs
Article #842Re: what to optimize with DAT drives?
From: Mark Simms
Date: Fri, 06 Nov 1992 13:29
Date: Fri, 06 Nov 1992 13:29
67 lines
3840 bytes
3840 bytes
>> To this end, I've been experimenting with compression during backup, >> and as I had already discovered compression and trying to stream the >> tape drive are not compatible. This depends a lot on the type of system you are running on. You give no indication of this in your posting so I have nothing to go on. If your CPU cannot compress data as fast as the tape drive can accept it, then the tape drive will not stream. This is a function of the transfer rate of the drive and the speed of the CPU. Other things can affect it such as the overheads involved in disc accesses, the tape driver and the tape drive itself. >> With this new DAT drive, is it as important to try to keep the drive >> streaming as it was with the 1/4" drive? DAT drives will wear out faster if they do not stream. The main failures are due to actual tape pulling time which affects head wear, repositions which affects wear in the tape pulling mechanism and tapeloading which affects wear in the tape load mechanism. If the tape does not stream, then this rapidly becomes the dominant factor. How significant it becomes, I do not know. >> As an example, I did a backup of a directory without compression >> that took 20 minutes to run, but kept the drive streaming (at least >> I assume it was streaming, since the activity light was flashing >> the whole time - the drive is so quiet, I can't *hear* it do >> anything). This is no indication that the drive is streaming. It is only an indication that the host is sending data to the drive. It is virtually impossible to tell if the drive is streaming. Even with the lid off you have to be very careful because you don't notice the reversals in motion. >> If I turned on compression, the backup took only 8 minutes total >> elapsed time, presumably since a lot less data had to be written to >> the tape (this particular directory average 70% compression), but the >> drive did not stream (again, as indicated by the activity light). Since the activity light stopped flashing, the drive was waiting for the host to send it more data. This would suggest that the limiting factor in the system is the time taken for the host CPU to compress the data. >> I checked the manual that came with the drive, and it doesn't seem >> to have anything in it pertaining to streaming and the life >> expectancy of the drive. There are life specifications, but I do not know whether they have been published or not. I would tend not to be too bothered about repositions. I do not think they have as big an effect as for QIC drives. I would suggest you consider an HP35480 DAT drive with built in hardware compression. This would take the load off the host and improve the performance of the system considerably. (It would also help pay my salary, so take my comments with that in mind :-) Mark Simms ---------------------------------------------------------------------- Opinions expressed are my own and are not intended to be an official statement by Hewlett-Packard Company ---------------------------------------------------------------------- Name: Mark Simms Profession: Software Engineer Organization: Hewlett-Packard Computer Peripherals Division Unix-mail: mark@hpcpbla.bri.hp.com ----------------------------------------------------------------------
Technical Details
Message-ID:
<10080017@hpcpbla.bri.hp.com>Path:
arc1-ca.pugleaf.net!sparky!uunet!charon.amdahl.com!pacbell.com!ames!saimiri.primate.wisc.edu!sdd.hp.com!scd.hp.com!hpscdm!hplextra!otter.hpl.hp.com!hpltoad!hpopd!hpcpbla!markReferences:
<4535@tiamat.fsc.com>Show Full Headers
Complete Article Headers
From: mark@hpcpbla.bri.hp.com (Mark Simms) Date: Fri, 6 Nov 1992 13:29:31 GMT Subject: Re: what to optimize with DAT drives? Message-ID: <10080017@hpcpbla.bri.hp.com> Organization: HP Computer Peripherals Bristol, UK Path: arc1-ca.pugleaf.net!sparky!uunet!charon.amdahl.com!pacbell.com!ames!saimiri.primate.wisc.edu!sdd.hp.com!scd.hp.com!hpscdm!hplextra!otter.hpl.hp.com!hpltoad!hpopd!hpcpbla!mark Newsgroups: comp.periphs References: <4535@tiamat.fsc.com> Lines: 67