steve@wolfen.cc.uow.oz (Steve Cliffe) (09/18/89)
Can someone please tell me why Sequent persists with the user-limit concept. We currently have a 2 processor symmetry with a 32 user limit. Pyramid dropped this quite a while ago and our Sun's have no such limit, so it is very hard for us to justify paying quite a lot of money for no physical gain. When we reach our user-limit response time is still quite good, so when our local sales rep asks us when we are going to upgrade our machine we have to say that we can't because our user limit won't let us get the most out of our current configuration so putting more hardware in is pointless. So, in the future we it comes to allocating courses to machines we will be forced to run more and more on our Sun's untill eventually, the Sequent becomes forgotten. Unhappy, Steve. Stephen Cliffe, | Phone: +61 42 270937 Dept. of Computing Science, | Fax: +61 42 297768 University of Wollongong, | ACSnet: steve@wolfen.cc.uow.oz Wollongong NSW 2500, | UUCP: ..!munnari!wolfen.oz!steve Australia. | Arpa: steve@wolfen.cc.uow.oz.AU
jim@cs.strath.ac.uk (Jim Reid) (10/01/89)
In article <6006@wolfen.cc.uow.oz> steve@wolfen.cc.uow.oz (Steve Cliffe) writes: >Can someone please tell me why Sequent persists with the user-limit >concept. This is because UNIX is ultimately licensed by AT&T. Sequent's agreement with AT&T means that they have to pay AT&T royalties based on the number of users of each system they sell. The N-user AT&T licences mean Sequent supply N-user limited systems. I find it distasteful that this policy means that university departments like ours have to pay enormous sums for a Sequent UNIX distribution that's binary only. It is all the more galling when AT&T will gladly let us have a source licence for the same machine for a few hundred dollars. [OK, AT&T don't provide Sequent source code, but Sequent could supply this for a nominal sum to people who had the right piece of paper from AT&T.] A better deal for academic DYNIX licensing is needed. How about it? Jim PS: I know Sequent will release source, but the price is not all that nominal. I suspect that the current conditions mean that a hefty sum for the N-user licence would still have to be paid on top.
perl@pbseps.UUCP (Richard Perlman) (10/01/89)
In article <6006@wolfen.cc.uow.oz> steve@wolfen.cc.uow.oz (Steve Cliffe) writes: >Can someone please tell me why Sequent persists with the user-limit concept > >Pyramid dropped this quite a while ago and our Sun's have no such >limit... I was told by Sequent that the user limit is part of the AT&T license agreement (I have not independantly verified this). We currently have a 32 user license, we can upgrade to 64 users or unlimited. I do not know why some other vendors do not use such a limit, but here is one possible way they could avoid it: Simply include a license sized appropiately to the machines maximum capacity. A large system like a Pyramid might simply charge you for the max license on purchase. A small system like the SUN 386i may include a license that it would be impractical to exceed based on system load/power. If you want more users on your Sequent, just buy a license for more users! I'd rather pay for what I need and know what I got. -- Richard Perlman * perl@pbseps.pacbell.com || {ames,sun,att}!pacbell!pbseps!perl 180 New Montgomery St. rm 602, San Francisco, CA 94105 |*| 1(415) 545-0233
csg@pyramid.pyramid.com (Carl S. Gutekunst) (10/02/89)
In article <633@pbseps.UUCP> perl@pbseps.PacBell.COM (Richard Perlman) writes: >I was told by Sequent that the user limit is part of the AT&T license >agreement.... I do not know why some other vendors do not use such a limit, >but here is one possible way they could avoid it: Simply include a license >sized appropiately to the machines maximum capacity. A large system like a >Pyramid might simply charge you for the max license on purchase. Both Pyramid and Sun used to have number-of-user licenses, since that was the way AT&T billed for it. But AT&T abadoned this licensing scheme years ago. Once AT&T dropped it, both Pyramid and Sun immediately dropped it too, since the customers didn't like it and it was a hassle to maintain. BTW, as far as I know, neither old OSx or SunOS had anything to enforce the user limit. <csg>
rsk@boulder.Colorado.EDU (Rich Kulawiec) (10/02/89)
In article <294@baird.cs.strath.ac.uk> jim@cs.strath.ac.uk writes: >I find it distasteful that this policy means that university departments >like ours have to pay enormous sums for a Sequent UNIX distribution >that's binary only. It is all the more galling when AT&T will gladly let us >have a source licence for the same machine for a few hundred dollars. I don't purport to speak for Sequent, but if I were in their shoes, I'd keep tight control on the parallel kernel sources. As far as I can tell, that's where most of their technology edge lies; it's my opinion that Sequent's kernel is the cleanest/most elegant/best multiprocessor Unix kernel on the market. (Yes, I've read the sources; the site at which I used to work had a Dynix source license and I signed the non-disclosure agreement.) I guess I see it as a key point of Sequent's competitiveness. One approach which will partially solve your problem would to buy an AT&T source license and then send it (with whatever fee it is these days) to Berkeley for a 4.3 BSD source tape. At the utility level, the BSD and Dynix distributions are reasonably close. I know that this isn't an all-encompassing solution to the problem, but it may help. ---Rsk
csg@pyramid.pyramid.com (Carl S. Gutekunst) (10/03/89)
>I don't purport to speak for Sequent, but if I were in their shoes, I'd >keep tight control on the parallel kernel sources. Actually, *any* kernel sources. There is a lot of value-added that can be done to the UNIX kernel. Stright multiprocessing isn't very special any more (and for that matter, Dynix isn't doing anything particularly unusual). But there is a lot of closely-guarded stuff regarding scheduling algorithms (for CPU, I/O, and Disk), processor affinity, distribution of interrupts, costs of context switches and cache/TLB flushes, and so on. Even SVR4 is a long way from what I'd call a "commercial" UNIX; you have to add a lot to it to make a production system. *That* is what gives companies their edge, and is why most companies are so protective of their sources. >it's my opinion that Sequent's kernel is the cleanest/most elegant/best >multiprocessor Unix kernel on the market. So how many *other* MP kernels have you seen? I suspect the UMAX and MACH people would love to discuss this with you. <csg>
pcg@thor.cs.aber.ac.uk (Piercarlo Grandi) (10/03/89)
In article <294@baird.cs.strath.ac.uk> jim@cs.strath.ac.uk (Jim Reid) writes: In article <6006@wolfen.cc.uow.oz> steve@wolfen.cc.uow.oz (Steve Cliffe) writes: >Can someone please tell me why Sequent persists with the user-limit >concept. This is because UNIX is ultimately licensed by AT&T. Sequent's agreement with AT&T means that they have to pay AT&T royalties based on the number of users of each system they sell. The N-user AT&T licences mean Sequent supply N-user limited systems. Bogus idea. It has been long since AT&T has gone to very cheap licensing; a 1-2 users royalty is $50 (fifty) and an unlimited users license is $150 (one hundred fifty). Royalties for older Unix versions (pre-system V) were both higher and more linked to # of users, but still per-user royalties were very small. It is simply that Sequent (like DEC for Ultrix, which used to charge a few thousand dollars for each block of eight users authorized) want to make money. You might want instead to investigate on getting MACH sources for the Sequent, especially if you are a University. Apparently, if you do it right, you can even get it outside the USA. Contact the MACH people at CMU! -- Piercarlo "Peter" Grandi | ARPA: pcg%cs.aber.ac.uk@nsfnet-relay.ac.uk Dept of CS, UCW Aberystwyth | UUCP: ...!mcvax!ukc!aber-cs!pcg Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg@cs.aber.ac.uk
jim@cs.strath.ac.uk (Jim Reid) (10/04/89)
In article <12288@boulder.Colorado.EDU> rsk@boulder.Colorado.EDU (Rich Kulawiec) writes: >In article <294@baird.cs.strath.ac.uk> jim@cs.strath.ac.uk writes: >>I find it distasteful that this policy means that university departments >>like ours have to pay enormous sums for a Sequent UNIX distribution >>that's binary only. It is all the more galling when AT&T will gladly let us >>have a source licence for the same machine for a few hundred dollars. > >One approach which will partially solve your problem would to buy an >AT&T source license and then send it (with whatever fee it is these days) >to Berkeley for a 4.3 BSD source tape. At the utility level, the BSD >and Dynix distributions are reasonably close. I know that this isn't >an all-encompassing solution to the problem, but it may help. The 'problem' is not the availability or non-availablity of Sequent source code. [We already have source code licences from AT&T and Berkeley anyway.] The problem is the anomalous way that licensing is handled. AT&T will give source licences to educational users for a few hundred dollars. The same educational users have to pay thousands of dollars in UNIX licence fees (the bulk of which will ultimately go to AT&T) when they buy a system like a Sequent. Why? It is a Good and Noble Thing that AT&T supply cheap educational licences for UNIX. It's a pity that UNIX vendors cannot do likewise, either because they don't want to or because their agreements with AT&T prevent it. Jim
prc@erbe.se (Robert Claeson) (10/04/89)
In article <86123@pyramid.pyramid.com> csg@pyramid.pyramid.com (Carl S. Gutekunst) writes: >Both Pyramid and Sun used to have number-of-user licenses, since that was the >way AT&T billed for it. But AT&T abadoned this licensing scheme years ago. What is their licensing scheme based on nowadays? -- Robert Claeson E-mail: rclaeson@erbe.se ERBE DATA AB
jdarcy@pinocchio.encore.com (Jeff d'Arcy) (10/04/89)
pcg@thor.cs.aber.ac.uk (Piercarlo Grandi): > authorized) want to make money. You might want instead to > investigate on getting MACH sources for the Sequent, especially On the other hand, if you don't want to fiddle with sources and want a good version of Mach running on a shared-memory symmetric multiprocessor, I have a suggestion for you... :-) (Check out the .sig if you don't get it) Jeff d'Arcy jdarcy@encore.com (508) 460-0500 Encore has provided the medium, but the message remains my own
mudd-j@giza.cis.ohio-state.edu (John R. Mudd) (10/05/89)
In article <PCG.89Oct3143413@thor.cs.aber.ac.uk> pcg@thor.cs.aber.ac.uk (Piercarlo Grandi) writes: >You might want instead to >investigate on getting MACH sources for the Sequent, especially >if you are a University. Apparently, if you do it right, you can >even get it outside the USA. Contact the MACH people at CMU! Incorrect. CMU, with the help of some Sequent developers, ported a version to the Balance some years ago, but it is an old kernel version and was never (for the most part) released externally. As far as I know, CMU has not done a port to the Symmetry, and given that mt. Xinu is taking over commercial distribution of Mach, I wouldn't expect such a thing to occur. The last time I talked to anybody at Sequent about Mach, the answer I got was that, although certain parts of Mach (like the VM system) would be explored, Sequent was quite happy with Dynix. I don't believe that Mach sources are allowed to be exportedbinaries I'm not sure about. Somebody from mt. Xinu can enlighten us to that regard. ... John ------------------------------------------------------------------------------ John R. Mudd mudd-j@cis.ohio-state.edu Department of Computer and Information Science, The Ohio State University 2036 Neil Avenue, Columbus, Ohio, USA 43210-1277