From owner-freebsd-hardware Mon Dec 30 11:29:42 1996 Return-Path: Received: (from root@localhost) by freefall.freebsd.org (8.8.4/8.8.4) id LAA12809 for hardware-outgoing; Mon, 30 Dec 1996 11:29:42 -0800 (PST) Received: from maelstrom.dial.pipex.net (maelstrom.dial.pipex.net [158.43.128.52]) by freefall.freebsd.org (8.8.4/8.8.4) with ESMTP id LAA12803 for ; Mon, 30 Dec 1996 11:29:40 -0800 (PST) Received: from solaat81.dial.pipex.com by maelstrom.dial.pipex.net (8.8.3/) id TAA19020; Mon, 30 Dec 1996 19:29:23 GMT Message-Id: <3.0.32.19961230192648.0068db48@pop.dial.pipex.com> X-Sender: aat81@pop.dial.pipex.com X-Mailer: Windows Eudora Pro Version 3.0 (32) Date: Mon, 30 Dec 1996 19:28:09 +0000 To: grog@lemis.de (Greg Lehey) From: Simon Reading Subject: Re: DAT reliability Cc: freebsd-hardware@freebsd.org Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: owner-hardware@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Greg, Thanks for the mail. Apologies if this reply is a little long. At 14:48 30/12/96 +0100, Greg Lehey wrote: >Simon Reading writes: >> 1. As you say, the SDT-7000 is a new design. Like ANY new product, I would >> expect more teething troubles than one which has been out in the market for >> longer. (NB. I'm not saying this was the cause of my problems). > >I don't think that *any* is valid. Many new products have teething >troubles; many others don't. During this discussion I have come to >the conclusion that the HP C1533A is a whole lot more reliable than >the 35480A, for example. Maybe ANY was a rash generalization ;-) >> 3. How many people listen to their DAT player four hours _every day_? How >> long would it last if they did? > >I listen to my CD player for hours every day. It's about 7 years old >now. I don't have a DAT player, but I would guess that the typical >duty cycle for hi fi equipment could be higher than for tape backup >devices. I've used personal cassette players (cheap, expensive, different brands) for three hours a day, on average, they lasted about a year. I have a personal CD now, so this is no longer a problem. I don't know about the duty cycle of DAT or HIFI. >>> I would guess that the 5200 has been >>> out for longer and that any bugs/problems would be more likely to be >>> observed/sorted out than any with the 7000. The small price difference >>> between the two models make me think that there has been little change in >>> the funamental mechanism design and that 8000rpm may be too fast to >>> transport the tape using the existing mechanism. >>> >>> The price of the unit has little or nothing to do with its cost, >>> design or taste when deep-fried. I've observed over the last few >>> years that as a general rule, most DAT units "just work". I've only >>> met a few "persistent plaintifs" who seem never to be able to get a >>> working unit. >> >> There are two separate issues here. >> 1. Infant mortality. The reason why I returned my SDT-7000 was because it >> didn't work. > >Agreed. But there was evidence in your case that the device had >already been installed somewhere and returned for some reason. True. >> The reason why I have not exchanged for another SDT-7000, is in case >> it is a problem with the batch. (I can't afford to waste time with >> another dud). Fait accompli. > >You're assuming that you would be better off with another brand. I >don't think that these problems extend to whole batches. >> 2. Expected Lifetime. As stated before, I'm much more interested in how >> long I could expect a DDS-2 to last. From correspondence I've received I'd >> guess 18 months+ light usage, six months or so heavy usage (I'm happy to be >> corrected on this). > >I suspect you're (mis)quoting me here. I was talking about the 35480A >when I mentioned 6 months. I'm pretty sure I've had the C1533 for >well over a year, and it has had well beyond the expected 12% duty >cycle in that time. These were intended to be ball park figures rather than direct quotes. In the interests of accuracy, and for reference, I have quoted mails which have been sent to me, below. David Dawes writes: >I've had a 35470A for about 3.5 years now, and haven't had any >real problems. I've had a few I/O errors recently, but nothing that >a pass of a cleaning tape doesn't fix. phr@netcom.com (Paul Rubin) writes: >I have a C1533a now a little over 1 year old. I've had no problems >with it but my usage has been pretty light. Any DAT drive will >wear out after a year or so of heavy usage. Al Dykes writes: >I run a shop that has 7 tape drives, 4 are HP-DAT and three are >Exabyte 8mm. The exabyte systems backup about 4GB every night. The >DAT drives do about 3GB. I do 100% verifies on about half my backups. >That of course doubles the # of hours on the drive. All of these >systems are at least 2 years old. The oldest is just about 3 years >old. >Until this week all the DAT drives were HP35480A OEM units. In my >experience DAT drives work for about 2 years of heavy use and die >suddenly of something that is clearly mechanical, generally eating a >tape in the process. I can live with this. Greg Lehey writes: >>> Warning: Quality control on these drives varies greatly. . Neither lasted more than >>> 5 months. > >Well, this sounds very much like a report from me. But two things >don't fit: > . . >2. I was talking about 37480As. These are an older version of DDS-1 > drive. I do have a C1533A, and so far I've had no trouble with it > (more than 6 months, anyway :-) I think, in fact, I've had it > about 15 months, and I do a nightly backup which usually fills a > tape. Rainer Vonsaleski writes: >I bought an HP C1533A in November 1995, and I have been very happy >with it. Initially, there were some problems getting the termination >right (My Micropolis 1G drive *looked* to be terminating the bus, >but it wasn't). None of the problems were the fault of the C1533A. > >The C1533A has a most unusually complete set of error statistic logs >available. The performance with 90-meter tapes (hey, I'm cheap) >is fantastic. I trust this thing! This is from someone who wants >a medium to transfer hundreds of reels of 9-track tapes to for >archival purposes. > >I've spun about 50 full tapes in 13 months. The C1533A has behaved >flawlessly. If I had the choice to make over today, I would choose it >again, even though the Sony is 33% faster (on paper). Why? (1) Most of >my surprises with HP over the years have been pleasant surprises. >(2) Sony is playing catch-up, and their new drive is still much too >new. > >If the Sony is giving your grief, I'd return it for the C1533A. >In a second. There were also been a few comp.periphs.scsi postings (which I have filed. . on a server which has gone down :-( ) Greg writes: > Simon Reading writes: >> Does one DDS-2 manufacturer produce more reliable drives than >> another? I don't think we're much closer to an answer on this >> question. > >There are bound to be differences in reliability. It's just the >question whether they are statistically relevant. True. This is necessarily a problem of considering anecdotal evidence. > My feeling is that >the technology has matured considerably in the last 5 years, and that >it will continue to mature. As a result, I would prefer a new model >over an old model, even if other factors (capacity, speed) remain the >same. Is it not true that existing products continue to be tweaked as problems are observed by users? - For instance on my car, the car locks were replaced with a new design following problems experienced in the cold. - A certain make of car engine had a revised valve gear following (unpublicized) user problems. - TV designs for the same model are updated as better components become available etc. Regards Simon