BOB%HOWARD%CENTER@CC.UTAH.EDU (08/04/87)
>We are looking to purchase some more disks for our VAXCluster. We, >in the past, have bought RA series disks. Now, SI makes MSCP disks that >are definitely much faster than RA disks. Does anyone have any experience >(positive or negative) with SI C series disks connected to HSCs? > We made the mistake of purchasing System Industry disks and have been paying for it over and over again. Our observations are that System Industry doesn't appear to take the time to design the things right in the first place and when they do break (not IF but WHEN), SI hasn't a clue what to do to fix them. More than once I've called them and they will come out, scratch their heads and say... "let's replace this board and then let us know if that works!" Great! What good is a disk if you can't trust the data on it. What's worse, the disks we have seem to corrupt or lose files even after they have passed the lame SI diagnostics. So when someone does discover that a file has been wrecked it actually occured months ago. Try sifting through months of backup tapes looking for the last version of the file that wasn't corrupt. Basically, we're stuck with the disks and we only use them for temporary stuff that we can afford to lose (read ... we don't use them). If you think that it's just that we got a lemon, think again. SI has replaced everything two or three times and they still won't work right. They don't have a clue what to do to fix them. When it came time for us to buy the last set of disks we did some serious analysis, SI was cheaper per megabyte than DEC but when you took into account the maintenance over five years DEC won! (Amazing isn't it) Anyway, Ours is not the first Hughes site to have problems with the disks. When I talked to other sites within Hughes they all agreed that SI is not worth getting unless it's the absolutly last thing you can do... and even then don't do it. The bottom line is that I wouldn't buy System Industry disks on a bet. If you really want to talk about how bad they are send me mail and we'll get on the phone together. (They are so dumb that they once (and only once) refered us to a potential customer as a reference after the disks had been up and down for six months and then couldn't figure out why the customer didn't buy their disks.) If you think I'm being hard on SI, give me a call, I've got a bunch of disks to sell you... real cheap. Bob Wheeler Howard Hughes Medical Institute
JNOLAN@IRLEARN.BITNET (John Nolan Computer Center UCD) (08/11/87)
I must take Bob Wheeler to task over his comments about SI disks. Maybe there is a big difference between SI USA and SI Europe but we look on System Industries as an integrity supplier. Fujitsu make all SI disks. We've had 3 eagles on a 9900 for over 4 years with no disk problems (two controller failures). We've recently put a C-Series Eagle on a HSC and it is fine. DEC over the last year had to replace all our RA81 HDA's (the glued v's screwed problem). An RA81 will only sustain about 22 I/O/sec and Eagle will do over 30 (RA81 seek/latency 28/8.3ms, eagle 18.7/6.3ms) SI told us there was a problem on earlier C-series disks due to a manufacturing fault by Fuji and they had to return 100's of disks, our salesman says the problem is now fixed (and from our C-Series disk I believe him). I wouldn't buy another RA81 (or even RA82) while I can get better performing, more reliable, cheaper and bigger disks from SI (the same goes for tape drives) John Nolan.
SYSRUTH@UTORPHYS.BITNET (08/12/87)
Below is an excerpt from a message that I sent to Bob Wheeler after his recent posting about his problems with SI disks. He suggested that I post it to the list as he has observed that not many pro responses make it. (He also said in his reply that he had written the original message the same night as yet another failure had occurred, which certainly explains why he was so emphatic!). So here it is. I hope that it is of some use to people. I have added a few details about the ages of the disks, and the controllers. Ruth Milner ----------------------------------------------------------------------- ... The only disks we have here which are non-SI are 4 RD53's. Every other disk we own (1 SI83A (also known as a Swallow or 9744), 16 Eagles, 1 ancient CDC 9775, 2 dual Eagles (9798's), 2 9784's, 1 9722, and 1 old CDC 9766 (removable pack)) has come through SI. On average we have less than 1 problem per month in total, counting all disks (and 9 controllers - 6 9900's and 3 QDA-50's). Once in a while a block goes bad; once in a *long* while a board goes bad; and only once, with just the 9775, did we have anything close to a head crash (it did require an HDA replacement, and we had one other 9775, which we have since sold, which had had 2 HDA replacements 18 months apart - they are *old* disks). The Eagles range in age from 3 months to about 5 years. The others (except of course the SI83A) are towards the high end of that, or older. Our service person is brilliant. The response time is generally very good, and he rarely has any trouble figuring out the problem. You did not mention what type of disks you got from SI. The SuperEagle made a lot of people mad (not surprisingly, although the problem there was more with Fujitsu than with SI; SI does not design all their own drives). But the fact remains that SI sells a lot of good technology and stands by it. If your service person is no good, complain loudly; presumably, if he/she is no good at all, there will be other customers in your area with the same feelings. If your disks have an unacceptable failure rate from day one, insist on replace- ments (as I seem to remember Kevin Carosso saying he did with his SuperEagles). But please, please, don't run down the entire company and every person in it as being beneath consideration for any product they carry. ... Ruth Milner Systems Manager University of Toronto Physics SYSRUTH@UTORPHYS (BITNET) ----------------------------------------------------------------------------- P.S. I did hear from Bob in more detail about what kinds of problems he is experiencing, and some of them are very strange (e.g. the same block assigned to more than 1 file, a disk coming up with the wrong device protection which can't be changed, and several others). They are using SIMACS to co-ordinate disk sharing, and some of the errors sound almost more like software problems. He has apparently had these escalated all the way up to the point where SI brings in the "gurus" from California, with no success in curing them. I am not familiar with what is required to make SI disks run on HSC's, but if it involves a modified DEC driver (as most MASSBUS and UNIBUS/QBUS SI disks do) then people should be aware that the local SI Field Service branch should be providing upgrade patches at each new version of VMS. Obviously these won't come when VMS does, since SI doesn't get it until we do, but if, a few months after your VMS arrives, you still haven't seen your SI patches, ask (although you shouldn't have to, at least not if you have any maintenance contracts).
OBERMAN@LLL-ICDC.ARPA ("Kevin Oberman, 422-6955, L-156") (08/13/87)
This is a response to Bob Wheeler's comment on System Industries disks. It sounds like he's had totally unacceptable support and an incredible number of problems. This is clearly NOT universal. I got my first SI disk system (9762 drive on a 9500 controller) back in 1975. While the controller has gone away for something a little closer to state of the art (and dual ported), the system never gave us any serious problems. The controller was still working fine the day it was retired. The drive is still working TODAY! That's 12 years. I currently have the following SI equipment: 5 9753 (Eagle) drives, 4 9766 drives (RM05 look alikes), 3 9762 drives (RM03s), 1 9733 (about 300 MB 8in.), 4 9763 (Super Eagles), 978x (160 MB Fixed). All are connected to 9900 controllers on either SBI or CMI interfaces. I have found a failure rate (except on the Super Eagles) of about 0. I would have to look at the recordsto find the last failure, but it was not this year. We have a very good serviceman. He more or less lives on site. (The above lists only my hardware. The site includes LOTS of other SI hardware; some systems are lots bigger than mine.) My success with SI has been outstanding! I still am buying SI disks. I just orders several SI83Cs. And I'm sure I'll have some SI93Cs soon. This mostly says: 1. Service depends on the competence of the local people. Bob's experiance tells me SI needs to work on the personnel in some of their fiel offices. 2. My experiance with hardware reliability is totally at odds with Bob's. I've always found SI hardware to be rock solid. I tend to wonder if something about Bob's environment is not to the liking of the SI equipment. Or maybe it's just bad service people who have compounded a single failure into a series of disasters. 3. Bob should complain LOUDLY to the senior folks in SIs service structure. I mean in Milpitas. And if that doesn't work, talk to the sales people. And if that doesn't help, talk to the district people. Sales can make all sorts of other people jump. And I would not put up with the problems Bob has had. R. Kevin Oberman Lawrence Livermore National Laboratory arpa: oberman@lll-icdc.arpa (415) 422-6955 Disclaimer: I an not endorsing any product. I work for Uncle Sam and to do so would make him upset. The opinions are mine and mine alone. I tend to ignore my boss most of the time anyway.
forrest@blia.BLI.COM (Jon Forrest) (08/13/87)
I'd like to relate my experiences with SI since they seem to be similar to Ruth Milner's. At my previous job one of the Vax's I managed had ONLY SI disks and tapes. In other words, NO DEC stuff. Needless to say, this made me a little nervous but this was the configuration when I took the job. Fortunateley, the system disk was a 9766 which is the same as a DEC RM05 so I could run the standard DEC DRDRIVER without patches and also use standalone backup without patches to recover any crashed system disks. Anyway, the SI salespeople were just the opposite as the DEC salespeople. The SI people would call you every week and tell you how much they miss you and how you're doing such a great job and if there's anything you need to just give them a call. The DEC people rarely returned calls and were generally all but impossible to reach. I guess that is what being on commission does to your sales force. Well, things were going pretty well at my site but at another site which I didn't manage they had a major disaster, loosing 4 9766's packs and heads. The repairperson thought he had fixed them only to find that he had screwed up and that the packs were being trashed again. As luck would have it, some of the packs had only been backed up to other packs so there was some risk that further disasters would destroy unrecoverable data. Needless to say, SI management was called and told that the repairperson would no longer be allowed at the site. Due to this disaster, and to the fact that SI was trying very hard to do a good job in this developing geographical area, the guy never showed up again. For a while they had a service manager person do the work. He was OK but it was clear that he couldn't do the work forever. So, the had to hire someone else. Here's where things got interesting. The guy they hired looked like a reject from the local Hell's Angel chapter. I'm not kidding. If you didn't know this guy you would not have let him into your site. The funny thing was that this guy could fix ANYTHING. I guess that there much be a similarity between disk/tape drives and Harley Davidsons. I never failed to be amazed when this guy would take the knife out of his boot to open envelopes. Or, when waiting for disks to format, he would break out the latest issue of Easy Rider magazine. Or, sometimes he would tell me about some of the biker bars he's been thrown out of. The point is none of this mattered because he fixed stuff. And it didn't break again. This, compared to the 3 piece bozo that had caused the disaster, was refreshing. I left this job about a year and a half ago but I've heard this guy is still there and everyone is happy with him. SI has given up trying to get this guy to wear a suit because he'd probably quit and then everyone would be pissed off at SI again. The moral of this story is that your success with SI will depend to a large extent on the quality of your repairperson. Of course, there may be inherent problems in the hardware or software you buy but that doesn't happen that often and, in the case of hardware, it probably isn't SI's fault since they don't make their own hardware. I did get tired patching DEC drivers and have since decided never to buy hardware from anybody that requires patching drivers. I might buy from SI again but I would check all their local references to see what they think of the local repair people. Also, if the quality of the repair people goes down I wouldn't be bashfull in letting management know about it. Jon Forrest forrest@blia.BLI.COM ucbvax!mtxinu!blia!forrest {pyramid|voder}!blia!forrest