jgd@rsiatl.UUCP (John G. DeArmond) (09/26/90)
In <LARRY.90Sep13141156@focsys.uucp> larry@focsys.uucp (Larry Williamson) writes: >I must write > 1 Gigabyte of data to a tape on a 386 system running >Unix 3.2. It may be either an ISA bus (or EISA bus system if a scsi >controller is available). I expect to use an Exabyte drive (what else >would be faster?), an adaptec 1542 controller (for both tape and >disk). >What data rate should I expect to be able to maintain for the duration >of the write? We have an exabyte drive on a Compaq 386/33 Deskpro using an Adaptec SCSI controller, a Maxtor 600 drive and ISC release 2.2. After my collegue Paul Anderson wrote a set of utilities to condition the drive and to guarantee correct blocking, thanks to ISC's crappy exabyte driver (full flame to follow), we see a write rate over the duration of the backup of between 600 and 700mb/hr. This is with all components right out of the box and no kernal tuning. We get about 400mb/hour backing up other machines over the ethernet using rsh. We've found a wide variety of video tapes to be reliable. IN particular the Sony P6-120mp metal particle tape ( ~$9.00 ea)appears to be the same tape as the $40.00 "certified" exatape. One thing to be aware of. There is a hole in the bottom of the cartridge between the spools but forward toward the tape. The tape drive contains a sensor that looks for reflective material in this hole. The certified tapes and the Sony plated metal tapes ( ~$14.00 ea) have a reflective foil in this hole. The P6-120 does not. In some versions of the drive, and in particular, the one OEM'd by NCR, the lack of this reflective foil reduces the tape capacity to about 200 mb. Insert some foil and you are off. The drive we bought directly from exabyte does not read this foil. I suspect that this is a strappable option, though I'm not sure becasue we did not get a hardware manual with the drive. The drive seems to be very reliable once one learns how to work with it. It is NOT like a DC-600-type drive. Blocking MATTERS. You MUST block your data on 1024 byte boundaries. And you MUST retension the tape before any opens. ISC apparently does not supply a utility; after an extensive RTFM, we wrote our own. And you MUST NOT attempt any operation until the green light is on. If you do, you must remove the tape and reinsert it (a 45 second operation.) About the only sour note has to do with Interactive Unix (as usual). Their driver appears to do little more than format SCSI commands for the drive. It appears to have no particular knowledge of the drive characteristics. It is in stark contrast to the exabyte driver supplied by IBM on the RS/6000. (Yeah, so I said something nice about the shitstation/6000. It won't happen again.) It took us several hours to figure out how to make a device for the tape. kconfig does NOT do it and the FM is confusing to a sin, mixing minor number information freely between QIC and exabyte type drives. And as usual, a call to tech support got a "Duuuuuuhhhh". To the point, we still have not figured out how to make a non-rewind device. We HAVE figured out how to make a kernal PANIC device. Just make a device, we call it /dev/xpanic, with a major number of 41, minor number 12. Then open the device and watch the machine panic and reboot! But once you tiptoe through the turdpatch, you'll find the exabyte about as fast a device as any backup unit you might find. John -- John De Armond, WD4OQC | We can no more blame our loss of freedom on congress Radiation Systems, Inc. | than we can prostitution on pimps. Both simply Atlanta, Ga | provide broker services for their customers. {emory,uunet}!rsiatl!jgd| - Dr. W Williams | **I am the NRA**