Date: Fri, 09 Jul 1999 01:39:57 -0700 From: The Clark Family <Clark@open.org> To: Greg Lehey <grog@lemis.com> Cc: jsd@gamespot.com, freebsd-questions@FreeBSD.ORG Subject: Re: HP T4000s Tape Drive problems Message-ID: <199907090813.BAA15329@opengovt.open.org> In-Reply-To: <19990709114248.R6035@freebie.lemis.com> References: <199907082224.PAA27583@opengovt.open.org> <199907080218.VAA14937@hostigos.otherwhen.com> <199907082054.NAA03455@hudsucker.gamespot.com> <199907082224.PAA27583@opengovt.open.org>
next in thread | previous in thread | raw e-mail | index | archive | help
>>> From what I remember: >> >> Order that the equipment came out (from my vantage point): >> >> DDS-1 >> DDS-2 >> Exabyte (DAT) > >Exabyte is 8mm, not DAT. > My mistake. I should have said 8mm. I was thinking about the distinction between the earlier exabyte formats, and the 8mm stuff. (I could be off on this too.) >> >> DDS-1 had one disadvantage, we couldn't run any of the drives for more than >> three years (nightly backups) without them wearing out. > >You were doing well. I never got one to survive for a year. I mis-stated this as well. The drives we were using were not DDS-1 (DDS-DC), they were only DDS. > >> DDS-2 was the tape format preferred by the UNIX heads at our shop. It was >> very similar to DDS-1 (used the same tape), but featured hardware based >> data compression. (4GB only) > >I'm not sure I understand this statement. You said above-- >correctly--that you could get (about) 4GB out of a DDS-1 drive with >compression (in fact, that would only be with 120m DDS-2 tapes; the >longest DDS-1 tapes were 90m and would give you about 2.5 GB >compressed). DDS-2 tapes were (are) only 120m, and they will give you >4GB uncompressed or about 8 GB compressed. > Another mistake. We were using DDS drives that did not have hardware compression. They had a 2GB native capacity with 90m tapes. They could not use 120m tapes. (The 120m tapes use different tape, not just more length. AFAIK.) We used software compression, and were able to get approx 4GB on one 90m tape. But DDS-1 drives had a native capacity of 2GB on a 90m tape? And with software compression, they should get 4GB (ideal). I could be wrong on this as well, but those numbers seem to go with the HP SureStore 5000 drives, and they are DDS-1 aren't they? >> On UNIX, where the backup software was the definition of primitive, >> hardware compression was the only type easily available. > >I don't understand this statement. On (the old version of HP) UNIX (we were using), where the backup software was the definition of primitive (compared to Arcserve on th e Netware servers (IMO)), hardware compression was the only (compression) type easily available ( at that time as far as I know). > >> The hardware compression, and the fact that the UNIX guys bought >> from higher priced VARs, made the DDS-2 stuff quite a bit more >> expensive. Maybe 1k$ (US) per drive. > >It's not the compression that made them more expensive. They were a >better technology, and lasted 18 months instead of 9. Again I made a mistake. The UNIX admins used a DCLZ capable version of the DDS drive. *Not the DDS-2 Unit.* 1k$ for a DDS-2 tape drive would make more sense. I think the DCLZ tape drive's data format would have been DDS-DC. They were able to get 4GB on a 90m tape without using software-based data compression. I think these drives had a dip switch that allowed the hardware compression to be turned on or off. When the compression was turned off, the DCLZ was essentially the same as the DDS (non-DDS-DC) model. > >> I seem to remember that Exabyte jumped into the fray at this point. (With 8mm.) > >Exabyte have been around for over 10 years, since before DDS. > But am I correct in remembering that they were working with a different tape technology for most of those 10 years? I seem to remember older exabyte tapes that looked like QIC. More like the old 1.2GB Tandberg stuff. (Magnus 1.2GB?) >> Their drives were based on 8-mil DAT technology, and because they >> had more surface area, they had a higher capacity. > >In fact, there isn't much in it. The 8202 would only give you about 2 GB. I didn't see exabyte stuff that early. I think the 8505 was the first exabyte 8mm stuff I saw. >> >> Second, is their speed. I don't remember the numbers off the top of my >> head, but 20MB / second seems to ring a bell. > >Nowhere near. From >http://online1.quantum.com/products/dlt/dlt4000/dlt_4000_features.htm: > Again, my mistake. I was thinking of the old DDS (pre DDS-DC) drives. At 180kB/Sec native, they would do about 10MB per *minute*. With a two to one compression ratio, we were seeing 20MB per *minute*. Thats what I was thinking of. > Features and Benefits > > High performance 1.5 MB per second native transfer rate So, if the drive can handle 1.5MB / sec continously, it should be able to move 90MB per minute onto tape. (Without compression.) With two to one hardware compression, 180MB per minute. So a 20GB native capacity tape should take about 4 hours to fill up. > > High capacity 20 GB native capacity > > High reliability and durability > > Backward compatibility to previous DLTtape formats > Speaking of which, have you seen any of the DLT2000 tape drives on the market? >> >> DLT does also have some pitfalls: >> >> One, is that they are expensive. I seem to remember that it was usual for >> the DLT4000 to go for about 4k$ when the first came out. > >The disk drive depot in Sunnyvale is currently selling refurb units >for $995. But the tapes are *really* *expensive*. I wouldn't be suprised to see them available refurbed pretty often. Quantum was suggesting that any drive that crashes with a tape inside be sent back to the factory for repair. Wouldn't these then become "refurbs"? As far as tape costs go: I think we were paying 80$ per (DLT) 20GB/40GB tape. 2$ per GB? I think we were paying 8$ per (DDS-1 90m) 2GB/4GB tape. 2$ per GB? I don't know what AIT or DDS-3 tapes cost, but when we made the transition to DLT, I don't think either of them were on the market yet. > >> Second, is that it takes quite a while for the tapes to load load and >> unload. This can be a problem in tape jukeboxes, or if you tend to need >> bits of data off the tapes often. > >They're still faster than Exabyte, and on par with DDS. > But slower loading than AIT? >Greg >-- Either I didn't get in the point about the drives being able to choke on a tape, or perhaps you didn't think it worthy of including in the reply. Do you think this is a non-issue? I've not had to remove a broken tape myself, and wouldn't be daunted by the task. But two of the administrators I work with have had to take turns pulling broken tape from the pick up reel, and were a little put off by the incident. A statistical anomaly? I don't know. Thanks for the info. [RC] To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199907090813.BAA15329>