Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!csd4.milw.wisc.edu!lll-winken!uunet!rick From: rick@uunet.UU.NET (Rick Adams) Newsgroups: comp.sys.sequent Subject: Re: Questions about Pyramid/Sequent Summary: an alternative opinion Message-ID: <52535@uunet.UU.NET> Date: 10 Apr 89 21:43:55 GMT References: <764@sactoh0.UUCP> <63900002@uxf.cso.uiuc.edu> Organization: UUNET Communications Services, Falls Church, VA Lines: 106 In article <63900002@uxf.cso.uiuc.edu>, bob@uxf.cso.uiuc.edu writes: > the support folks at sequent are reasonable people who treat any problem you > encounter as THEIR problem and resolution is quick. Sorry, I can't let this go by unchallenged. The support people at Sequent are well meaning and try hard, but... In several of my dealings with them, they do not believe me when I say there is a problem and IF I can convince them that a problem exists, they MAY take the trouble to fix it. (I think the problem is not convincing the first level support people, but in getting whoever in engineering has the responsibility to accept that there might be a problem). Let's take a few examples: There is a bug in their X.25 software that leaves a process running and a user "logged in" after the connection has closed. The current "resolution" of this as I understand it (it's been a while) is that its not a bug. Well, since Dynix counts the number of users currently logged in and limits that number to whatever license size you bought, it's clearly a denial of service problem. (I still claim it's a bug in the X.25 driver) Reloading the X.25 software fixes things (and logs out anyone else who happens to be logged in). There is a bug somewhere in their TCP or IP such that when subjected to extremely high loads (Imagine what the SMTP load on uunet is when it's been off the Internet for a few days and suddenly 10,000 machines want to send it mail.) it "loses" SMTP connections requests. By lose I mean that there is a daemon running, listening on the port, but if you try and connect (even through the software loopback) you get NEITHER a connection accepted nor rejected. Its like the host is down. They don't believe this is a problem since they can't reproduce it in their "lab". Fortunately, this is a fairly rare problem, because I don't expect it to ever get fixed. Currently, we have a very serious problem with terminal ports hanging. This hang means that the Sequent totally ignores data sent from the terminal. Nothing short of rebooting will clear this hang. It's as if the driver were stuck in some flow control state that it can't be freed from. (The Sequent sends to the terminal fine). We have to reboot the system on a daily basis to free stuck ports. Obviously not a good situation for a production system. I'm not sure they believe this is a real problem, but it sure is for us. (We will ignore the kludgely "modem control" on their Systech boards. Same ugliness as Suns onthe ALM1) [This one should be easy to fix. It happens several times per day.] The basic problem with the Sequent product is that they have truly wonderful hardware and barely adequate software. People are buying their systems despite the software. Sequent is still, for the most part, shipping 4.2bsd (bugs and all). Their System 5 compatibility is not quite that old (4.2bsd is 7 years old now), but is only V.2 and is not very current. Sun is shipping current versions of Berkeley and ATT code. This is proof that you CAN do it despite claims about going through channels, quality control, etc. If a company the size of Sun (or even DEC) can ship the 1986 4.3BSD release, surely a smaller company like Sequent can ship it within 3 YEARS of release. Since this started as a Pyramid vs Sequent discussion, I should point out that Pyramid software is YEARS ahead of Sequent. I have no direct experience with Pyramid hardware (donations accepted...), but Pyramid software is clearly far superior to Sequents (despite the dual-universe abomination). Maintenance is an interesting issue. Sequent's hardware is so reliable and their software support so marginal that it is extremely tempting to do without a maintenance contract at all. (Yes, it's time for me to renew the maintenance contract...) Every week or two, someone who is considering buying a Sequent asks me how I like my Sequent. I keep telling them the same thing. The hardware's great, the software sucks. If Sequent would take half the time they spend screwing around searching for a better TP1 benchmark and put it into cleaning up the basic system (Just try and stay within 1 or 2 years of your competitors guys. I'm not asking for parity) they could have a truly great product. Judging from the people I've talked to, I bet they lose 10 sales a year because of software inadequacy. You'd think that would justify 1 single engineer to take programs off of the latest distribution tapes and compile them for the next distribution. I KNOW thats all it takes, since virtually all of the programs I use on a regular basis, I TOOK OFF THOSE TAPES. I've replaced all of the communications/networking programs. Thats why the system I have is usable. (If you're a Sequent user, get the berkeley networking tape. So far, all the programs I have tried compile and run with no problems.) The truly maddening part of all this is I bet I've talked to 25 different Sequent people in detail (including the president of the company, who I had dinner with at the last Sequent Users group meeting). Every one of them seemed very interested and concerned and wanted to help. But, if I were grading them it would have to be A for effort and C+ for product. I'm hoping for a major miracle in the next software release. I'm expecting, well... ---rick P.S. Overall, I don't think Sequent is really much better or worse than most other computer companies (Although, it's software is clearly behind). I don't want anyone to mistakenly believe that Sequent is the fantasy vendor that some people have been portraying.