ken@sitename (Ken Seefried iii) (05/22/91)
----- I could use some advice on choosing a backup device... I've finally reached the point where the multi-GB nightly back- up's are taxing our Exabyte and my nerves. Therefore, I'm going to be going to multiple drives to handle the load. Now, the easy solution would have been to buy another Exabyte or three, but that would be too simple... Seems that one of our kernel jocks has a pal at <unamed mini-super- computer> company in the quality assurance department that claims that Exabytes just aren't the way to go...that, indeed, the DAT is the One True Backup Device (tm), and that the Archive Python is the One True Incarnation of said Device. Sigh... So I come to the net hoping someone has done the definative comparison of backup device technology and can lay down the pros and cons of each device, given that it will be used in a heavily networked, *very* heterogenous environment, and needs to be able to work with a wide variety of Unix machines (we have a habit of moving the things around). Many, many thanks... -- ken seefried iii "I'll have what the gentleman ken@dali.cc.gatech.edu on the floor is having..."
cmf851@anu.oz.au (Albert Langer) (05/22/91)
In article <29543@hydra.gatech.EDU> ken@dali.cc.gatech.edu (Ken Seefried iii) writes: >So I come to the net hoping someone has done the definative comparison >of backup device technology and can lay down the pros and cons of >each device, given that it will be used in a heavily networked, *very* >heterogenous environment, and needs to be able to work with a wide >variety of Unix machines (we have a habit of moving the things around). I can't offer a comparison of the technology (and those I have read I find no more convincing than those arguing for the superiority of VHS or BETAmax video formats). I went for DAT for the same reason I went for VHS as a videotape format - because I thought other people who I will want to exchange tapes with are more likely to end up with DAT. The reason I think that is because DAT is based on a consumer audio technique which has been held off the market in the USA and Australia for spurious copyright reasons but is already popular in Japan and can be expected to become popular in other countries too. When that happens I would expect the cost of consumer audio DAT to plummet a year or two after the initial "prestige" introduction and the cost of scsi DAT to follow it downwards. Then with DAT dramatically cheaper than Exabyte etc I would expect it to become the "standard" on unix boxes, regardless of technical merit. Only argument I can see against this reasoning is that there is now a competing consumer audio digital audio tape format, I think called DCC and I think based on cassettes the same physical shape as existing audio cassettes (and with player/recorders designed to handle both). If that format became dominant then the same argument would apply to it, but as far as I know it isn't available for computer backup now anyway. -- Opinions disclaimed (Authoritative answer from opinion server) Header reply address wrong. Use cmf851@csc2.anu.edu.au
wadswort@babss.UUCP (John H. Wadsworth) (05/22/91)
In article <29543@hydra.gatech.EDU> ken@dali.cc.gatech.edu (Ken Seefried iii) writes: >----- > ....deleted text about ..to buy more exabytes, that is the question.... >Seems that one of our kernel jocks has a pal at <unamed mini-super- >computer> company in the quality assurance department that claims that >Exabytes just aren't the way to go...that, indeed, the DAT is the One >True Backup Device (tm), and that the Archive Python is the One True >Incarnation of said Device. > I have no experience with DAT and could only repeat net.rumors so I can't tell you if they are better or worse. I do have experience with exabytes. Multiflow Computer was a beta test site. The first revs of exabytes were pretty bad. Some days they worked, some days they didn't. Exabyte has made some changes in their firmware and testing procedures. They have improved about a zillion percent since I started working with them. THIS IS THE PART YOU NEED TO KNOW...... There should be a revision sticker on the drive. Look for the line that lists the mx board. The prom should be 4$24 or above (like 4$25 4$25a ). 4$24 made a big improvement in the "works one day but not the next" and 4$25 seems to extend the mtbf but I don't know why. Before the 4$24 firmware i used to swear some drives were haunted. That's why I called them hexabytes :-)
carroll@ssc-vax (Jeff Carroll) (05/24/91)
In article <29543@hydra.gatech.EDU> ken@dali.cc.gatech.edu (Ken Seefried iii) writes: > >So I come to the net hoping someone has done the definative comparison >of backup device technology and can lay down the pros and cons of >each device, given that it will be used in a heavily networked, *very* >heterogenous environment, and needs to be able to work with a wide >variety of Unix machines (we have a habit of moving the things around). I've found Exabytes _qua_ Exabytes to be unreliable and unfixable. On the other hand, I've used third party backup units that seem to hold up fairly well. If you buy from an outfit that provides good field service for them, Exabyte tape servers are OK. I've had good luck with Micro Technology (though they tend to be VMS-centric). -- Jeff Carroll carroll@ssc-vax.boeing.com "...and of their daughters it is written, 'Cursed be he who lies with any manner of animal.'" - Talmud
mark@hpcpbla.HP.COM (Mark Simms) (05/24/91)
Unfortunately, the DAT/8mm discussion is a bit of a religious one with very little rationality and even less data. I work for Hewlett-Packard. I am involved in the development of HP's 3.5" form factor DAT drives. Therefore I am biased. If you find your Exabyte reliable and has suitable performance, then you should probably stick to it. Changing to DAT would involve a whole new integration cycle before you could get the system working as you like it. It may require new drivers, new software and so on to support DAT drives. You may need to attach the DAT drives to a different system in order to get them to work. Having said that, how do the technologies compare? The 8mm format has a higher capacity per tape and a higher transfer rate. The new 5Gbyte drives are impressive on paper, but I have seen very little of them in practice. 8mm is a single vendor technology. 8mm has 5.25" full height form factor that cannot be easily shrunk. 8mm is not supported by any major computer manufacturers as a long term standard for data interchange. The DAT format has a lower capacity than the 8mm format although the use of 90m tapes and on drive data compression will help remedy this in future. DAT is shipped by at least three different vendors using different mechanism suppliers. HP and archive have 3.5" drives in evaluation quantities now. HP and DEC already support the DAT DDS format and have a long term commitment to it. HP and Archive have announced drives that support 90m tapes that bring the capacity close to that of the standard 8mm model. HP has also announced a drive with on drive data compression which may improve the capacity more. I cannot recall whether Archive have announced a DC product or not. I would expect to see these drives available some time this summer. The current HP DAT drive is probably more reliable than the Archive drive. It can get more data onto the tape due to requiring fewer data rewrites. It is full height 5.25" form factor while the archive drive is half height. The HP drive is a bit of a bus hog. (This is fixed in the new HP drive that is comparable to the new Archive drive.) The main reasons for switching to DAT are for long term support and data interchange. You may get gains in reliability. If you have, or are looking into getting, some systems that require half height or 3.5" drives, then you should consider DAT. Probably the most important factor when choosing a backup system is the amount of support you get from the vendor. If a system vendor has a standard backup system, then use it. It will probably be the most reliable. If not, then find someone who will sell you a complete solution including support. It may cost a little more, but you are more likely to get a system that works. 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 Occupation: Research and Development Organization: Hewlett-Packard Computer Peripherals Division Unix-mail: mark%hpcpbla@hplb.hpl.hp.com Address: Filton Road, Bristol BS12 6QZ, United Kingdom Phone: +44-272-799910x22174 Fax: +44-272-236091 ----------------------------------------------------------------------
mbl900@anusf.anu.edu.au (Mathew BM LIM) (05/28/91)
I am new to this discussion so I may have missed the answer to the following question. One of our vendors has told me that the 5GByte figure on the capacity on the newer Exabyte drives is achived by on board data compression and that the actual capacity depends on the compressability of your data. So presumably the actual capacity of a tape is anywhere from 2.3GBytes to 5GBytes. Is this true? If it is, does anyone have any figures on what the "typical" capacity is when backing up a "typical" filesystem? Thanks -- Mathew Lim, Unix Systems Programmer, ANU Supercomputer Facility, Australian National University, GPO Box 4, Canberra City, ACT, Australia 2601. Telephone : +61 6 249 2750 | ACSnet : Mathew.BM.Lim@anu.oz Fax : +61 6 247 3425 | Internet : Mathew.BM.Lim@anu.edu.au
dtb@adpplz.UUCP (Tom Beach) (05/29/91)
In article <1991May28.130950@anusf.anu.edu.au>, mbl900@anusf.anu.edu.au (Mathew BM LIM) writes: > One of our vendors has told me that the 5GByte figure on the capacity on the > newer Exabyte drives is achived by on board data compression and that the actual > capacity depends on the compressability of your data. So presumably the > actual capacity of a tape is anywhere from 2.3GBytes to 5GBytes. Is this > true? If it is, does anyone have any figures on what the "typical" capacity > is when backing up a "typical" filesystem? The new 5 GB Exabyte is REAL capacity e.g. not compressed. The new DATs however DO use hardware compression to achieve 4 GB. The new 4 GB DATs are compressed with a HP algorithm based on LZ-2 data compression. These drives presume a 2x compression from your actual data. For most files this is in fact conservative but if you have many compressed files already in your filesystem problems arise. Tom ------------------------------------------------------------------------ | Tom Beach : Sr Project Engineer : Mass Storage Technology | | phone : (503) 294-1541 | | email : uunet : dtb@adpplz.uucp | | ADP Dealer Services, ADP Plaza, 2525 S.W. 1st Ave, Portland OR, 97201 | ------------------------------------------------------------------------
rlr@alumni.colorado.edu (Roger Rose) (05/29/91)
> One of our vendors has told me that the 5GByte figure on the capacity > on the newer Exabyte drives is achived by on board data compression and > that the actual capacity depends on the compressability of your data. Well, your vendor is simply wrong. ;-) There are 3-rd party compressed versions of the EXB-8200 (2.4Gb) drive available; however, the EXB-8500 (5Gb) drive is not compressed. The capacity gain was accomplished by doubling the track density. A compressed version of the drive has also been announced, but the compression will be in addition to the **native** 5Gb capacity. > So presumably the actual capacity of a tape is anywhere from 2.3GBytes > to 5GBytes. Is this true? If it is, does anyone have any figures on > what the "typical" capacity is when backing up a "typical" filesystem? This depends on a few factors. Two major factors being how good the error statistics are on the tape and how well you keep the drive streaming. The stated capacity is the formatted physical capacity of the drive. (You can get the exact figures to logical-end-of-tape from the manuals. The numbers to physical-end-of-tape vary slightly from tape to tape.) Tape quality affects capacity, because questionable blocks are rewritten and this consumes space. Actual loss on data-grade tapes should run less than 1% even after the tapes have been used several times. Streaming affects capacity, because the drive attempts to maintain streaming by inserting gaps into the data. In order to stream an EXB-8500, you need to maintain a 500Kb **average** data rate from the host. If this data rate is not maintained, then gaps are inserted whenever the drive's cache empties. (There are ways of tuning the drive to minimize the impact of transfer rates, if your driver software knows how.) -- Roger Rose {rlr@boulder.colorado.edu}
se@IKP.Uni-Koeln.DE (Stefan Esser) (05/29/91)
In article <1991May28.130950@anusf.anu.edu.au>, mbl900@anusf.anu.edu.au (Mathew BM LIM) writes: ||> One of our vendors has told me that the 5GByte figure on the capacity on the |> newer Exabyte drives is achived by on board data compression and that the actual |> capacity depends on the compressability of your data. So presumably the |> actual capacity of a tape is anywhere from 2.3GBytes to 5GBytes. Is this |> true? If it is, does anyone have any figures on what the "typical" capacity |> is when backing up a "typical" filesystem? | NO, its definitly NOT TRUE, that the 5GB Exabyte drives use data compression to achive the higher capacity ! The Exabyte 8500 has twice as many data heads as the 8200 (8200: 2 data + 1 servo, 8500: 4 data + 1 servo). Transfer rate and capacity are doubled, independent of your data. There will soon be DAT drives with extended capacity using 50% longer tapes and data compression technology. Stefan Esser -- Stefan Esser, Institute of Nuclear Physics, University of Cologne, Germany se@IKP.Uni-Koeln.DE [134.95.192.50]
tim@gumby.cs.caltech.edu (Timothy L. Kay) (05/29/91)
mbl900@anusf.anu.edu.au (Mathew BM LIM) writes: >I am new to this discussion so I may have missed the answer to the following >question. >One of our vendors has told me that the 5GByte figure on the capacity on the >newer Exabyte drives is achived by on board data compression and that the actual >capacity depends on the compressability of your data. So presumably the >actual capacity of a tape is anywhere from 2.3GBytes to 5GBytes. Is this >true? If it is, does anyone have any figures on what the "typical" capacity >is when backing up a "typical" filesystem? Your vendor is confused. It is true that Exabyte now offers compression capability with their EXB-8200-SX unit, and that will allow you to store approximately 5 GB on a tape (assuming 2X compression). HOWEVER, Exabyte ALSO has an EXB-8500 which stores 5 GB UNCOMPRESSED. You can get that with a compression chip (or compress the data on the host, and then you'll be able to put 10 GB on a tape (again assuming 2X compression). If your vendor is saying that you should get a DAT, consider that it is quite easy these days to buy a 2.3GB EXB-8200 drive (bare) for about $1800. I just hooked one up to my Iris 4D-80GTB trivially, and it is working nicely. If you ask via email, I'll send you the name of the vendor. Tim
dnichols@ceilidh.beartrack.com (DoN Nichols) (05/29/91)
In article <9850023@hpcpbla.HP.COM> mark@hpcpbla.HP.COM (Mark Simms) writes: [ ... ] >The DAT format has a lower capacity than the 8mm format although the use >of 90m tapes and on drive data compression will help remedy this in >future. DAT is shipped by at least three different vendors using This isn't one of my usual newsgroups, so this may have already been hashed out without my knowledge. My question concerns what happens when compression is in use, given that the drives become much less efficient when a block of a length not reasonably matched to the size of a single diagonal track (esp one just slightly larger) is written, and more data is not available at the time the drive writes the track. It would seem to me that the compression is likely to provide unpredictable length blocks of data being fed to the tape for fixed-length blocks coming from the system. How is the system to keep the blocksize optimum, especially if the data is coming from another system over ethernet? (Since delays in ethernet transfers may keep another block from being ready soon enough to use the rest of that track.) Thanks DoN. -- Donald Nichols (DoN.) | Voice (Days): (703) 664-1585 D&D Data | Voice (Eves): (703) 938-4564 Disclaimer: from here - None | Email: <dnichols@ceilidh.beartrack.com> --- Black Holes are where God is dividing by zero ---
rlr@alumni.colorado.edu (Roger Rose) (05/30/91)
> ... It is true that Exabyte now offers > compression capability with their EXB-8200-SX unit,... Actually, the EXB-8200-SX unit is the high speed search version of the EXB-8200. It offers no native compression facilities. I'm not aware whether third party compression options are available for the SX. -- Roger Rose {rlr@boulder.colorado.edu}
bovet@hao.hao.ucar.edu (Ray Bovet) (05/31/91)
In article <1991May28.130950@anusf.anu.edu.au> mbl900@anusf.anu.edu.au (Mathew BM LIM) writes: >I am new to this discussion so I may have missed the answer to the following >question. > >One of our vendors has told me that the 5GByte figure on the capacity on the >newer Exabyte drives is achived by on board data compression and that the actual >capacity depends on the compressability of your data. So presumably the >actual capacity of a tape is anywhere from 2.3GBytes to 5GBytes. Is this >true? If it is, does anyone have any figures on what the "typical" capacity >is when backing up a "typical" filesystem? > There are two major flavors of Exabyte drives. The original drives are the model 8200. These drives hold up to 2.3 Gbytes on a standard 106 m (P120) 8mm video tape. The "new" Exabyte drive is the EXB8500. This drive doubles the track density so that a 106 m tape can hold about 4.6 Gbytes. Exabyte claims 5 Gbytes on the 8500 because they now sell 112 meter tapes for it. All of these numbers are without any compression. Several vendors of Exabytes have incorporated their own compression schemes external to the drive. In addition, Exabyte is committed to providing internal compression as an option at some future date. The compression you achieve will depend greatly on the data you are storing. This means that you won't have an easy way of knowing ahead of time how much data you can safely put onto a single tape. In my mind this is a pretty unpleasant situation to be in. Fortuantely for us, the 5 Gby capacity of the EXB8500 drives seems adequate for our needs. Ray
caserta@athena.mit.edu (Francesco Caserta) (06/03/91)
In article <1991May28.223259.149101@rrz.uni-koeln.de>, se@IKP.Uni-Koeln.DE (Stefan Esser) writes: |> |> There will soon be DAT drives with extended capacity using 50% longer |> tapes and data compression technology. |> |> Could you provide us with more info, please? Thanks, Francesco Caserta
dwells@fits.cx.nrao.edu (Don Wells) (06/05/91)
In article <1991Jun3.013930.23211@athena.mit.edu> caserta@athena.mit.edu (Francesco Caserta) writes: In article <1991May28.223259.149101@rrz.uni-koeln.de>, se@IKP.Uni-Koeln.DE (Stefan Esser) writes: |> There will soon be DAT drives with extended capacity using 50% longer |> tapes and data compression technology. Could you provide us with more info, please? Kimberley Electronics, located in New Jersey, advertises in the back pages of magazines like Digital Review and Digital News. They are currently offering Wangtek DAT drives which have firmware support for 90-meter tapes, which have 2.0_GB capacity. Their price for a SCSI DAT in a box with power supply and connectors is US$1500. Compression will soon be offered if it isn't already available. -- Donald C. Wells Associate Scientist dwells@nrao.edu National Radio Astronomy Observatory +1-804-296-0277 Edgemont Road Fax= +1-804-296-0278 Charlottesville, Virginia 22903-2475 USA 78:31.1W, 38:02.2N