fpt@wgivax.UUCP (02/18/86)
Since SUN has become almost a household word in the UNIX community,
I thought I would tell the net about our recent (somewhat rough)
experiences as a SUN customer.
My hope is that the right people at SUN may see this, and that some
of our questions and problems will be addressed. Certainly, current
and potential customers of SUN deserve to know.
I am forwarding a copy of a memo from one of us. I have added my own
comments in places (in parens).
MEMO
from: Don Mosley 2-17-86
Washburn Direct Marketing
to: Concerned parties
subject: Sun Microsystems
This memo is an effort to vent some of the frustrations accumulated
in the acquisition of a Sun3/180. Some of the problems described here have
been fixed, most are still pending resolution.
1. Delivery
System ordered 10-5-85, partially delivered 2-12-86. Somewhat
understandable; Sun was moving to a bigger plant. The irritating
part is that delivery was promised on 1-24, 1-31, 2-5, 2-10.
(Actually, the first promised delivery was '30 days', approx. 11-5)
2. Incomplete
System was ordered with 16 line ALM and color monitor. Monitor is
promised to ship 2-28. No one knows when the ALM will ship. The
portion that did arrive is missing the software installation manual.
The Ethernet transceiver kit was missing the screws to hold cable tap
into transceiver body. CPU carton did not have a packing list.
(ALM is a 16 port asynch. mux. The most aggravating thing to me was
that it was not until we got the box, looked inside, 'hey, where's
the mux?', that sun told us 'available in limited quantities
starting in April'. We bought the machine to use. We needed it in
the promised thirty days. Now we find out we can't put serially
connected users on it until April?)
3. Lack of Communication
We attempted to purchase a GCR 1/2 inch tape controller for the Sun3.
It took 4 weeks for the salesman to give us a part number and price
for the parts. Order was placed 1-28. We found out on 2-10 that the
tape controller was not available from Sun as a spare part or as a
system option. Sun office in Raleigh states the reason for this as
"they can't figure out how to add the part number to their spare parts
database". Really inspires confidence in this organization.
(Anyone know how to get 9 track tape on a Sun 3? Can we buy the
parts from someone else? We don't need a drive, just controller.)
The Sun offices in Raleigh and Atlanta have repeatedly failed to return
my phone calls. When they do return calls, the answer is frequently
"I'll look into it and call you right back", which rarely happens.
Sales related questions to Sun in California are referred to Raleigh.
4. Overpricing
The above mentioned tape controller, a Xylogics 472, is reported to
cost $3,390 from Sun. The same part from a computer systems retailer
costs $1,100. The 380 mb SMD disk subsystem from Sun costs $19,900.
The same parts, Xylogics 450 controller, Fujitsu Eagle, VME to Multibus
adapter would cost $10,750 on the open market.
5. Questionable Engineering Practices
The above mentioned disk controller is designed to operate 4 SMD drives.
As supplied by Sun it will run only 2. The reason for this is that
Sun takes the industry standard socket connectors on the Xylogics board
and jumpers them out to D-connectors on the back panel. This
arrangement only has room for 2 drives to connect.
To go along with this arrangement, each drive will require an adapter
panel to convert from the D-connectors back to socket connectors that
most SMD drives use. Why not just leave the socket connectors on the
controller alone? That would allow the use of 4 drives on one
controller and would eliminate a needless link of cabling which is a
potential point of failure.
I feel certain that the Sun3 will be a tremendous system once we get
it all together. I would also recommend strongly that we look at other
vendors before purchasing any more computers from Sun. Surely no company could
stay in business and treat it's customers as bad as Sun has.
END OF FORWARDED MEMO
It's seldom that one sees bad press on Sun. I'm wondering what you all think.
Are we unique? Is our problem that we only bought 1 instead of 10? Are these
problems only happening in the East? Or in the Atlanta district (includes
Raleigh)?
Any comments will be appreciated.
Fred Toth
Don Mosley
Washburn Graphics, Inc.
Charlotte, NC
decvax!mcnc!unccvax!wgivax!{fpt,dm}roy@phri.UUCP (Roy Smith) (02/20/86)
In article <147@wgivax.UUCP> fpt@wgivax.UUCP writes: > The above mentioned disk controller is designed to operate 4 SMD drives. > As supplied by Sun it will run only 2. The reason for this is that > Sun takes the industry standard socket connectors on the Xylogics board > and jumpers them out to D-connectors on the back panel. This > arrangement only has room for 2 drives to connect. Far be it from me to defend Sun against bad press (and yes, I agree that their peripherals are priced out of this world), but the problem with the disk attachments may be more due to FCC rules than any fault of Sun's engineering. As of about 2 years ago, the FCC has strict rules on RFI emissions from computers which are creating packaging headaches for manufacturers. Pre-FCC, nobody paid much attention to packaging; all you needed was a 19-inch rack to hold everything off the floor and enough switched outlets to plug everything into. Side panels and fans helped move cooling air around, but other than that it was mostly just decoration. You used to be able to simply run flat cables from the controller, out the back of the CPU box, and across the floor to the drive. Now you now have to go through shielded bulkheads and use shielded external cables. There is extra cost involved, and a lot of re-thinking in the packaging department. Why do you think DEC's MSCP cables cost so much? All that shielded coax with those crazy bulkhead connectors ain't cheap. I't hard to see how a regular flat cable connector could meet the FCC requirements. It may be a pain, but in the long run I guess it's all for the best. I'm sure the radio astronomers and hams appreciate the effort. -- Roy Smith, {allegra,philabs}!phri!roy System Administrator, Public Health Research Institute 455 First Avenue, New York, NY 10016
thomas@utah-gr.UUCP (Spencer W. Thomas) (02/20/86)
I hate to be a pessimist, but our experience is that MOST small computer
companies treat their customers "this way". Delays in delivery,
non-availability of new hardware, and so on seem to be the norm in this
industry.
--
=Spencer ({ihnp4,decvax}!utah-cs!thomas, thomas@utah-cs.ARPA)pavlov@hscfvax.UUCP (840033@G.Pavlov) (02/21/86)
Re the question whether other people having same experience: for the first 3 paragraphs, I thought that your letter was coming from someone I just talked to in Cambridge. order date, 'promise' dates, and arrival dates virtually identical. His is a 3/160, tho. It's too bad. I really hope Sun gets on track and succeeds. But they're going for broke with sales, and can't keep up with the growth.
mp@allegra.UUCP (Mark Plotnick) (02/21/86)
Your experiences are not unique; we've been having problems with Sun for as long as anyone here can remember (I've only been dealing with them for the past 18 months). I know they're still considered by many to be a startup company, so perhaps some of our problems are unavoidable. We think the machines are really nice - our lab, 100 computer scientists, has 75 of them, with 15 more on order - but the sales and support leave much to be desired. Sales: We and other customers at our site have had to wait up to 3 months to get quotes on certain items (disk cables, source code, maintenance contracts). As with you, California says they can't help us and we have to wait for the local office to get around to doing things. We don't appreciate the fact that, after we've placed orders, they can rarely tell us when the orders will be shipped. Memory boards and color upgrades have typically taken 6 to 8 months to ship, and we're still waiting for some things we ordered last summer. YOU try telling a department head who wants to put a color monitor on his Sun that you have no idea when it will come, and that it may not be until 1987! When ordering entire systems, one gets faster delivery (within a month or two of the projected date). But file servers always seem to get delivered last, and in some cases (as with the people in a neighboring building who have had diskless Sun-3 clients sitting unused since late December) this makes the portion of the order that's been delivered pretty useless. We've gotten into the habit of stating "no partials" on our orders. Maintenance: We purchased an "on-site" service contract in late 1984, and were told in writing that it was Sun's intent to put someone physically at our site full-time. We paid a lot for this (1.3% of list price per machine per month), yet they still don't have someone residing on-site. I was told by someone in another company that Sun's service contract prices were lowered in March 1985; when I asked Sun about this, they said that price list didn't apply to us, since we were paying extra for the COMMITMENT to have someone resident on-site. After a series of meetings over the last few months, Sun agreed to lower the prices to those in the July 1985 price list. We've also heard that a $1500/site/month software contract has been available for some time, possibly dating from early 1985. Our lab is paying over $3000 per month for software maintenance (assuming $70 to $75 per machine per month, which can be extracted from the different rates in the current price list), and there are other customers at this location that are also paying this rate. We have asked Sun several times to give us retroactive credit, since we certainly would have converted to the lower-priced contract had we known about it, but we're still waiting for action. The skill of the field service people has varied widely, and I could tell some real horror stories, but don't think it's worth describing them here because I feel it's mainly a result of Sun's growing pains. I'll note a few things: (1) some people will still not let any Sun FE touch their systems because of some particularly bad experiences. (2) Sun's on-site service contract differs from many other companies' in that it doesn't specify the maximum time period that can elapse between the logging of a call and the point at which an FE comes to see you. It can sometimes be more than a week. We received some Sun-3's in late December/early January. None of them are yet in service, and the warranties are half over. 2 of the 7 systems were short-shipped (missing cables), and repeated phone calls have not yet gotten any action. We don't have a vme tape drive, and thus can't load the 3.0 software onto the Sun-3's (the version of the software that will work on our Sun-2 systems, which is where all our tape drives lie, is not available to us yet). Sun couldn't accommodate us in any way (e.g. by loaning us a vme-to-multibus adapter or a vme tape drive for a day). Last week, we got tired of waiting for a miracle and got things loaded onto one system by hauling its disks over to Sun's district office and loading the software there ourselves. We brought the disks back here, put them in a 160, and the 160 dropped dead after 2 days (bad cpu board). The person logging the service call here thought that maybe Sun could bring in a new cpu board when they dropped by to fix some other systems (we have about 50 Sun-2 systems on contract, and the FE comes here at least once a week). NO! We had to buy an on-site service contract for this new system first! After some people higher up at Sun were contacted, they said they could make an exception in just this one case (but we had already decided to call up the eastern service center, which arranged to have California ship us a cpu board in the mail with hopefully a week's delay); the effort expended in getting this concession underscored to us that Sun has little interest in building good working relationships with their customers. Despite the fact that my lab has 1.5 million dollars worth of Suns, and spends over $100K per year for maintenance, Sun won't do anything they're not required to do. Sun may argue that they can't make exceptions if they are to make money; we argue that buying computer systems is more than just signing contracts and exchanging cash for hardware, and that when our users and staff get a negative feeling about a vendor, it's time to start looking for another vendor. Mark Plotnick allegra!mp
grandi@noao.UUCP (Steve Grandi) (02/22/86)
I guess it is nice to know that our bad experiences are not unique!
But it sure is frustrating to have equipment arrive that doesn't work,
to have to deal with seemingly incompetent salespeople, and to have
delivery dates treated like fiction. None of these things are
disasters; we now know how to make a mousehouse work and when
something is REALLY wrong, we get one day service on replacement boards.
And we chose Suns over microVAXes, Integrated Solutions boxes, et al.
with our eyes open.
Two specific questions: Has anyone yet received a 68020 upgrade for
their VME Sun-2s? Is the cgi library source included on your source
distribution tape for v2.0?
--
Steve Grandi, National Optical Astronomy Observatories, Tucson, AZ, 602-325-9228
{arizona,decvax,hao,ihnp4,seismo}!noao!grandi grandi%draco@Hamlet.Caltech.Edumark@umcp-cs.UUCP (Mark Weiser) (02/23/86)
In article <455@carina.noao.UUCP> grandi@noao.UUCP writes: >Is the cgi library source included on your source >distribution tape for v2.0? Only the source to cgitool came, not libcgi or libcgi77, with our 2.0 source tape. -mark -- Spoken: Mark Weiser ARPA: mark@maryland Phone: +1-301-454-7817 CSNet: mark@umcp-cs UUCP: {seismo,allegra}!umcp-cs!mark USPS: Computer Science Dept., University of Maryland, College Park, MD 20742
freed@aum.UUCP (Erik Freed) (02/24/86)
I won't bother to quote the original article. I have also had some
bad experiences with Sun as well as good. The bad seems to be that
they are in their corporate expansion, very likely to cut some
corners especially as far as making the waiting customer happy.
We were told that they would make a certain shipping date on a
*very* important shipment and about one week before the expected
arrival of equipment I called about something else and was told
that they had rescheduled delivery a month later and had neglected
to tell me! I have yet to find out whether they will make good on this
because it was equipment necessary to make a super important show.
They knew how important it was because we had negotiated the OEM
agreement based on their delivery as scheduled. So at least
verbally it was a breach of contract. I think that they will make good
on it, but I am waiting.
As far as the mark-up of peripherals that is actually pretty
standard (look at DEC prices). I feel that Sun in general is probably
not so bad. Sun is trying to become another DEC and their success
in the cutthroat Generic Unix box is probably going to be related to
how well they keep their delivery schedules (and Q/C)
On the other hand alot of their support has been good and I am certainly
very happy with the product.
--
-------------------------------------------------------------------------------
Erik James Freed
Aurora Systems
San Francisco, CA
{dual,ptsfa}!aum!freeddoug@isieng.UUCP (Doug Swartz) (02/26/86)
From: Doug Swartz I am the President of Integrated Solutions, a company that competes directly with Sun Microsystems in the 4.2 BSD workstation business and am thus clearly NOT an unbiased observer. I noticed your net mail with great interest. I am not sending this mail out as sour grapes nor as an effort to sell you anything, so I am not going to tell you at all about our product. Further, before I begin I do not mean this mail to imply that our products are perfect nor that we have only completely happy customers. I liken what has happened with Sun Microsystems to be very much like the case of Emperors clothes. They seem to have a very effective mechanism (which I admire by the way) for suppressing all negative information about themselves and their products. By talking to the people I know that work at Sun, I have found that they have had two layoffs, had or have ( I get two different stories) acknowledged mix and match problems between their CPU and memory boards, ie, memory boards have to be hand selected to work with CPU cards, had problems some time ago with defective memory chips and have the same quality control problems that all fast growing American companies seem to have. And yet not of these things ever appear in the press even though Sun appears to make the front page of Computer Systems News at least every other issue. It seems that everyone, including the press, feels that everyone else feels that Sun is such a great company that no one dares to criticize it. At first I felt guilty about writing this mail because I feel that it kind of a perversion of what is supposed to be a network for distributing technical information to bad mouth a competitor. However, I remember numerous articles extolling the virtues of the Sun product written both by current and former Sun employees and I didn't feel guilty any more. I am glad that you had the courage to say "Hey that guy's at least a little bit naked". On your question about the SMD cables, it is actually technically a violation of the SMD specification to have any connectors in the link between the SMD controller and the drive and can lead to higher than normal soft error rates although the SMD specification is so conservative that I doubt you will ever see this problem unless your cables are extremely long.