Interest In NTI/48.2 Project
June 1996
Stuart Herbert (S.Herbert@sheffield.ac.uk)Document copyright ©. All rights reserved.
Abstract
JISC, as part of its New Technologies Initiative, has funded the
University of Sheffield to supply and support Generic NQS, a batch
processing system, to UK Higher Education. This work is now coming
to a close, and this paper presents the results of a survey of UK HE
sites.
Contents
Click here for a plain-text version of this paper. Click here for a copy of this document in Microsoft RTF format, suitable for printing (if available).
Executive Summary
Executive Summary
- There are at least 41 installations of Generic NQS which have
submitted documentary evidence to us, including 35 UK HE sites.
- There are a further 15 probable installations of Generic NQS at
UK sites, and a further 80 probable installations of Generic NQS
outside the UK.
- 59% of installations of Generic NQS are across two or more
operating systems at any one site.
- No correspondants are willing to pay current UK academic pricing
for commercial batch processing systems; indeed, 16 correspondants
indicated that they were unwilling to pay at all.
- 100% of respondants who expressed an opinion stated that they were
happy (or better) with both the work carried out under grant
NTI/48.2 and the support provided under grant NTI/48.2.
However, the lack of detailed documentation was mentioned as an
important omission.
- 92% of correspondants who expressed an opinion stated that they
wished the work on Generic NQS to continue beyond the end of the
JISC-funded period; only 8% of correspondants did not wish the
work to continue.
- 71% of correspondants who expressed an opinion stated that they
were willing to contribute time and effort (mainly through beta
testing) to help support future work on Generic NQS. 29% of
correspondants were unable, or unwilling, to contribute.
Introduction
Purpose And Scope
The purpose of this paper is to present the results of a survey
carried out amongst UK Higher Educational sites and users of Generic
NQS, and to draw some comparison between these results and previous
surveys.
The scope of this paper is NOT limited to usage within UK Higher
Education; Generic NQS is used world-wide, and feedback from the
Generic NQS community at large is important for future directions of
the maintenance of Generic NQS.
Methods
A questionnaire [1] was drawn up, and sent by electronic email to
- NQS-Announce@mailbase.ac.uk
- SGI-Systems@mailbase.ac.uk
- UCISA-Directors@mailbase.ac.uk
A copy was also sent to Tom Franklin, the JISC NTI Co-ordinator.
Retured questionnaires were examined, and their data transfered into
the tables which form later sections of this paper.
This data, where possible, was then compared with data collected
earlier during the project [3], [4], and conclusions drawn.
Data Collected
Introduction
In this section, we will present the data collected from the returned
questionnaires, to allow for an analysis in a later section.
Table 1: Replies Received
> ==================================================================
> Reply From | Email | Type
> ==================================================================
> Aarchen University of | Thomas Eifert | D
> Technology | (Eifert@rz.rwth-aachen.de |
> Computing Centre | |
> ------------------------------------------------------------------
> Bombardier AeroSpace Group | Stephen Rompala | D
> | Steve.Rompala@mfg.canadair |
> | .ca |
> ------------------------------------------------------------------
> Freie Universitaet Berlin | Holger Busse | C
> Institut fuer Physik. und | busse@chemie.fu-berlin.de |
> Theroret. Chemie | |
> ------------------------------------------------------------------
> German Aerospace Research | Eberhard Mater | D
> Establishment | rzgj@dlrvms.do.dlr.de |
> ------------------------------------------------------------------
> Institut de Pharmacologie et | Emmanuel Courcelle | D
> de Biologie Structurale - | manu@ecstasy.cemes.fr |
> I.P.B.S.(ex-L.P.T.F.)/C.N.R.S | |
> ------------------------------------------------------------------
> London School of Economics | Sujit Choudhury | C
> IT Services | S.Choudhury@lse.ac.uk |
> ------------------------------------------------------------------
> SGS Thomson Microelectronics | Richard Wignall | C
> | richardw@bristol.st.com |
> ------------------------------------------------------------------
> University of Birmingham | John Newbury | C
> Information Services | J.P.Newbury@bham.ac.uk |
> ------------------------------------------------------------------
> University of Cambridge | Dr. G. Carver | B
> Centre For Atmospheric Science| (glenn@atm.ch.cam.ac.uk) |
> ------------------------------------------------------------------
> University of Cambridge | Dr. Tim Perkins | B
> Department of Pharmacology | (Tim.Perkins@ddg.phar.cam |
> | .ac.uk) |
> ------------------------------------------------------------------
> University of East Anglia | Dr. Andrew Boswell | D
> Computing Centre | A.Boswell@uea.ac.uk |
> ------------------------------------------------------------------
> University of Edinburgh | Keith Farvis | C
> Computing Services | keith@ucs.ed.ac.uk |
> ------------------------------------------------------------------
> University of Exeter | Phil Chambers | D
> IT Services | P.A.Chambers@exeter.ac.uk |
> ------------------------------------------------------------------
> University of Leeds | Timothy Nicholson | B
> Polymer Science & Technology | phy6tmn@irc.leeds.ac.uk |
> ------------------------------------------------------------------
> University of Leeds | Jasbinder Singh | C
> University Computing Service | ecl6jasb@cif.leeds.ac.uk |
> ------------------------------------------------------------------
> University of Leicester | Dr. T. A. Claxton | A
> Chemistry | (tac@leicester.ac.uk) |
> ------------------------------------------------------------------
> University of Leicester | Dr Lee Hallam | D
> Computer Centre | (ldh2@le.ac.uk) |
> ------------------------------------------------------------------
> University of New South Wales | Michael Shepard | B
> Australia | (M.Shephard@unsw.edu.au) |
> School of Chemistry | |
> ------------------------------------------------------------------
> University of Oxford | John Peden | C
> Molecular Biology Data Centre | (johnp@molbiol.ox.ac.uk) |
> ------------------------------------------------------------------
> Institut fuer Physikalische | Bernd Kallies | D
> und Theoretische Chemie, | (kallies@serv.chem. |
> Universitaet Potsdam | uni-potsdam.de) |
> ------------------------------------------------------------------
> University of Salford | H.D.Ellison@ais.salford | D
> Academic Information Services | .ac.uk |
> ------------------------------------------------------------------
> University of Sheffield | Mark Taylor | C
> MAC Services | (M.Taylor@sheffield.ac.uk) |
> ------------------------------------------------------------------
> University of Sheffield | Linda Gray | C
> Automatic Control & Systems | linda@acse.shef.ac.uk |
> Engineering | |
> ------------------------------------------------------------------
> University of Ulster | R.Wilson@ulst.ac.uk | D
> System Services | |
> ------------------------------------------------------------------
> University of Wales, | Robert Ash | D
> Aberystwyth | rwa@aber.ac.uk |
> Computer Unit | |
> ------------------------------------------------------------------
> University of Wales, Bangor | Dafydd Roberts | D
> Computing Laboratory | (D.Roberts@bangor.ac.uk) |
> ------------------------------------------------------------------
> University of Wales, Cardiff | Andrew Cormack | C
> Computing Service | (cormack@cardiff.ac.uk) |
> ------------------------------------------------------------------
> University of York | Dave Atkin | D
> Computing Service | dlal@unix.york.ac.uk |
> ------------------------------------------------------------------
> Universitaet Zuerich | Ulrich Berhard | C
> | (rzubu@rzu.unizh.ch)
> ------------------------------------------------------------------
> Unknown | Rafael Pappalardo | A
> | (rafapa@mozart.us.es) |
> ------------------------------------------------------------------
Key For Type Of Reply Received
> ------------------
> A : Self
> B : Research Group
> C : Department
> D : Site
> ------------------
Summation
- 30 replies were received, 19 from UK HE sites.
- 2 replies were on behalf of the individual.
- 4 replies were on behalf of groups within a department.
- 11 replies were on behalf of departments within an organisation.
- 13 replies were on behalf of organisations.
Table 2 : Batch Systems Used
> ==================================================================
> Batch System | Platform(s) | Happy/Unhappy
> ==================================================================
> CERN NQS | AIX/IRIX/SunOS| Happy
> | Solaris |
> ------------------------------------------------------------------
> CERN NQS | AIX/SunOS | Fairly happy except for lack of
> | Solaris | support for AIX4.
> ------------------------------------------------------------------
> DQS | IRIX, OSF1, | Mostly happy but there are major
> | SunOS | problems with DQS: Undefined limit
> | | of total jobs in queue, poor
> | | scheduling control (doesn't reorder
> | | queue correctly), poor reliability
> | | (sometimes requires a qstat to
> | | notice that jobs are present.
> | | However, can be made to work in
> | | small group (12 machines) environ.
> ------------------------------------------------------------------
> DQS + in-house| Digital UNIX | Following our own customisation,
> enhancements | | we have a stable queueing system
> | | which meets the needs of our users
> | | and our requirements to load
> | | balance the machines in the cluster.
> ------------------------------------------------------------------
> Fujitsu NQS | UXPM, UXPP, | No opinion expressed
> | UXPV |
> ------------------------------------------------------------------
> Generic NQS | OSF/1 | Happy
> ------------------------------------------------------------------
> Generic NQS | AIX, IRIX, | Happy: our needs are well met; it's
> | UXPV, HP-UX | available on all our platforms.
> ------------------------------------------------------------------
> Generic NQS | HP-UX, Linux, | Happy. It is very stable and does
> | IRIX | a very good job with respect of
> | | load balancing our workstation
> | | cluster
> ------------------------------------------------------------------
> Generic NQS | IRIX, HP-UX | Happy. Satisfies our basic
> | | queueing needs. Possibly room for
> | | a NQS-lite configuration option
> | | which would give you the batch
> | | queues and queue manipulation
> | | options but cuts down on control
> | | of resources by NQS and even the
> | | networking aspects of NQS.
> ------------------------------------------------------------------
> Generic NQS | AIX/IRIX/SunOS| Happy
> | Solaris |
> ------------------------------------------------------------------
> Generic NQS | HP-UX, Linux | Happy
> ------------------------------------------------------------------
> Generic NQS | Solaris, IRIX | Happy
> ------------------------------------------------------------------
> Generic NQS | AIX 3&4/HP-UX | Very happy
> ------------------------------------------------------------------
> Generic NQS | Solaris | Happy
> ------------------------------------------------------------------
> Generic NQS | Solaris | So far GNQS seems to be the right
> | | batch system for our need.
> ------------------------------------------------------------------
> Generic NQS | IRIX 5.3 & | Happy
> | 6.0.1 |
> ------------------------------------------------------------------
> Generic NQS | IRIX, AIX | Happy
> ------------------------------------------------------------------
> Generic NQS | ULTRIX, OSF1 | We are quite happy with NQS when
> | SGI | it runs on a machine standalone.
> | | But we found it rather difficult
> | | to install it for running of
> | | several machines.
> ------------------------------------------------------------------
> Generic NQS | IRIX 5.3 | Has proved very useful in the
> | | couple of months we have been
> | | using it. Together with some
> | | shell scripts I wrote allows easy
> | | use of unused processing capacity
> | | across cluster.
> ------------------------------------------------------------------
> Generic NQS | Digital UNIX | I am very happy with the
> | | development and support supplied
> | | by Univ. Sheffield via Stuart.
> ------------------------------------------------------------------
> Generic NQS | IRIX | Happy - easy to install, configure;
> | | Source available (for small
> | | changes); runs on SGI IRIX; uses
> | | the features of the IRIX OS (pset)
> ------------------------------------------------------------------
> Generic NQS | SunOS | Happy
> ------------------------------------------------------------------
> Gould Batch/ | SunOS, ULTRIX | Happy
> Tape | Digital UNIX |
> ------------------------------------------------------------------
> Loadleveller | AIX/SunOS | Not happy with IBM support and
> | Solaris | the tool is too complex.
> ------------------------------------------------------------------
> Loadleveller | AIX | I cannot answer this question
> ------------------------------------------------------------------
> LSF | OSF/1 | Happy. Does all main things
> | | required. Would be even better
> | | if we were using it on systems
> | | that could (securely) share files,
> | | load balance, etc. Well supported.
> | | No obvious hooks to allow local
> | | extension though.
> ------------------------------------------------------------------
> SGI 4D-NQS | IRIX | Happy - it is reliable but
> | | unsophisticated
> ------------------------------------------------------------------
> SGI 4D-NQS | IRIX | Reasonably happy, but intend to
> | | upgrade to Generic NQS prior to
> | | upgrading from IRIX 5.3 to 6.x.
> ------------------------------------------------------------------
> SGI 4D-NQS | IRIX | Happy (does what we want)
> ------------------------------------------------------------------
Summation
- There are 29 batch systems installed amongst the respondants.
- 8 different products are used.
- 13 different operating systems are used.
- 17 installations are across two or more operating systems.
- 1 installation is less than satisfactory. The other 28 state
that they are happy even if they go on to describe problems or
that in-house modifications were required in order to make the
product viable.
Table 3 : Reasons For No Batch System Usage
> ==================================================================
> Reason Given
> ==================================================================
> People seem satisfied with raw UNIX facilities.
> ------------------------------------------------------------------
> To my knowledge there is no use made of NQS at Salford.
> ------------------------------------------------------------------
> Cost & Availability (currently evaluating Generic NQS)
> ------------------------------------------------------------------
> Does not exactly fit our requirements, we have been interested in
> LSF which covers interactive use as well - but have generally
> tried to work round having to use this additional level of
> technology.
> ------------------------------------------------------------------
> Low resources
> ------------------------------------------------------------------
Summation
- 5 respondants did not use any batch processing at all.
Table 4 : Purchasing Commercial Systems
> ==================================================================
> Pay 15,000 pounds for | Willing To Pay ...
> site-license? |
> ==================================================================
> No | User demand probably not sufficient to
> | justify paid-for software. We have major
> | problems in persuading users to submit
> | jobs to a batch queue, rather than simply
> | backgrounding them under Unix.
> ------------------------------------------------------------------
> No | Computer Centre responsibility
> ------------------------------------------------------------------
> No | This question is somewhat too late since
> | we have GNQS ;-)
> ------------------------------------------------------------------
> No | I feel too GNUish to pay for software
> ------------------------------------------------------------------
> No | Preferably as little as possible ;-)
> | Depends on the license, but probably
> | 400US total for our three computers
> ------------------------------------------------------------------
> No | If possible, nothing
> ------------------------------------------------------------------
> No | Very little! have always had them free in
> | the past! say 100 pds p.a.
> ------------------------------------------------------------------
> No | Don't have that kind of money for a system
> | that could operate from free software.
> | Say 1000-2000
> ------------------------------------------------------------------
> No | 0
> ------------------------------------------------------------------
> Absolutely not - not | I would need a very good reason to buy
> while good high- | such a commercial package. My upper limit
> quality freely- | would be ~1000 (with no recurrent charges)
> available NQS | but I can currently see no reason to buy
> alternatives are | such a package whilst software like GNQS
> offered that do all | is available.
> that I want. |
> ------------------------------------------------------------------
> No | Hard to say - budget available would
> | depend on more than one division, and
> | would require investigation.
> ------------------------------------------------------------------
> Very unlikely | It's not my decision, but as Generic NQS
> | is free to UK-HE, my guess would be
> | nothing ;-)
> ------------------------------------------------------------------
> No | 5,000 max.
> ------------------------------------------------------------------
> No. | ?
> ------------------------------------------------------------------
> No | I can not say about this at present. Since
> | we will be using GNQS only on one machine,
> | it seems too high a price to pay.
> ------------------------------------------------------------------
> No | Maybe 1,000
> ------------------------------------------------------------------
> No | Given Monsanto-NQS is already running
> | quite satisfactorily, we would not pay
> | for an alternative.
> ------------------------------------------------------------------
> No | Up to 100 stlg per machine
> ------------------------------------------------------------------
> No | Nothing
> ------------------------------------------------------------------
> Not unless this was an| No information given.
> absolute requirement |
> of ours, which it is |
> not. |
> ------------------------------------------------------------------
> No | Nothing - we use free software because
> | we need it across several platforms,
> | don't want to pay recurrent costs, and
> | consider access to the source code to be
> | essential.
> ------------------------------------------------------------------
> No - would make do | Probably just a nominal sum
> with own scripts |
> ------------------------------------------------------------------
> No | 200-300 pounds at most
> ------------------------------------------------------------------
> Probably not. | When we were evaluating queueing systems
> | we considered up to 5K sterling for a
> | commercial system.
> ------------------------------------------------------------------
> No | As little as poss.
> ------------------------------------------------------------------
> No | Nil
> ------------------------------------------------------------------
> No | Maybe 1,000 pounds
> ------------------------------------------------------------------
- All 27 respondants who expressed an opinion stated that they
would not pay 15,000 pounds sterling for a site-license for
a commercial batch processing system.
- 16 respondants indicated that they were unwilling to pay for
a batch processing system.
Table 5 : Satisfaction With NTI/48.2 Work
> ==================================================================
> Aware | Happy With ...
> Of Work | Work? | Support?
> ==================================================================
> Yes | Yes. | Yes. All queries/bug reports
> | | have been answered very
> | | promptly.
> ------------------------------------------------------------------
> Yes | A great job has been | Support was immediate and
> | done to NQS to make it | competent - what else could
> | available on even more | we ask for?
> | platforms |
> ------------------------------------------------------------------
> Yes | Yes, fast and | Should be some better
> | competent | support for HP-UX, but this
> | | is my problem
> ------------------------------------------------------------------
> Yes | Yes. Very useful | Our needs are quite well
> | central site for the | satisfied by NQS as it
> | latest versions of NQS | stands, so we haven't
> | bug reports and fixes | required much by way of
> | and general info on | support. But the NQS
> | configuring and | mailing-list and the
> | running an NQS system | WWW site have proved very
> | | useful at times.
> ------------------------------------------------------------------
> Yes | GNQS is a stable running | Stu Herbert had always
> | system. It is easy to | open ears for proposals
> | install, to handle and to | of users in order to
> | maintain. We decided for | improve details of the
> | this system because the | system.
> | support and improvement of |
> | it seemed to be continuous. |
> | so we thought its usage |
> | would be an investment for |
> | the future. We tested other |
> | batch queueing systems very |
> | carefully (Cern NQS, DQS) & |
> | were surprised about the |
> | fact that only GNQS |
> | supported near all of our |
> | requirements. |
> ------------------------------------------------------------------
> No (1) |
> ------------------------------------------------------------------
> Yes | Can't say yet; still getting| Yes. I've had fast email
> | GNQS to run on all platforms| response to my problems
> | However, if it works as | from Stuart Herbert which
> | promised, it should be an | have (nearly) all been
> | excellent product ideally | solved. The email lists
> | suited to our needs. | are useful and the ftp/www
> | | sites are fast.
> ------------------------------------------------------------------
> Yes | Yes. The install process | Yes, the minor errors found
> | is vastly superior. The new| has been corrected almost
> | features are really useful | immediately. Some
> | | commercial houses should
> | | learn from this.
> ------------------------------------------------------------------
> Yes (2) | Yes, the work has been | Yes, the level of support
> | excellent. From my | has been excellent. Stuart
> | experience with CERN and | has maintained a good
> | Monsanto-NQS, I would say | turnaround time on the
> | that Stuart has done an | messages to the NQS mailing
> | excellent job of tidying up | lists. He has also shown
> | the code & the installation | enthusiasm for fixing bugs
> | of NQS. I do regret that | when reported and has been
> | he was unable to complete | very helpful when I have
> | the documentation however. | sent in queries about NQS.
> | This was a problem with the | I shall regret Stuart's
> | other versions of NQS and | departure.
> | remains a problem with GNQS.|
> ------------------------------------------------------------------
> Yes | The work carried out has | Yes, the level of support
> | been very well done. NQS has| is again very well done
> | helped us better admin our | indeed.
> | user environments. |
> ------------------------------------------------------------------
> Yes | We have just started testing| Yes - the WWW pages are
> | the upgrade from 4D-NQS to | very clearly presented,
> | Generic NQS, and it looks | and the information seems
> | very promising so far. | comprehensive.
> ------------------------------------------------------------------
> Yes | N.A. | Yes
> ------------------------------------------------------------------
> Yes | Not investigated | No details given.
> ------------------------------------------------------------------
> Yes | It was publicised and was | Yes.
> | targeted at the right people|
> | Papers comparing various |
> | batch systems were well |
> | written. |
> ------------------------------------------------------------------
> Yes | N.A. | N.A.
> ------------------------------------------------------------------
> Yes | Yes | Yes, some early problems
> | | were sorted out very
> | | quickly. Suggestions of
> | | mine foor solutions were
> | | incorporated in the fixes.
> ------------------------------------------------------------------
> Yes | Yes | Don't know
> ------------------------------------------------------------------
> Yes | Yes, no details yet, because| Yes, good WWW presentation
> | of very new GNQS user | and good support.
> ------------------------------------------------------------------
> Yes | I have not followed it too | We have not used the
> | closely. | software in service.
> ------------------------------------------------------------------
> Yes (3) | Yes. | Yes
> ------------------------------------------------------------------
> Yes | Very happy - it provides us | Response to questions has
> | with some useful management | been very welcomed
> | tools |
> ------------------------------------------------------------------
> Yes | The product has been very | Yes ... the level of support
> | well researched and | is very good and comparable
> | developed considering the | with the DEC product support
> | size of the development | (and they only support a
> | team. | fraction of the operating
> | | systems and platforms)
> ------------------------------------------------------------------
> Yes | Yes. I have downloaded the | Yes, the support was much
> | NQS and compiled for SunOS, | better than expected. My
> | Solaris and AIX4 with few | email questions had a
> | problems | response time of < 1
> | | working day. Also the
> | | response fixed problems.
> ------------------------------------------------------------------
> Yes | Yes! GNQS is easy to install| Yes. With the mailing lists
> | and is moving on. | there are a lot of hints and
> | | tips. It is easy to report
> | | bugs and problems and I
> | | get immediate answers to
> | | such reports.
> ------------------------------------------------------------------
> Yes (4) | N.A. | N.A.
> ------------------------------------------------------------------
> Yes. | Yes - a lot of effort | N.A.
> | appears to have gone into |
> | the project. We were |
> | regularly updated on |
> | available versions. |
> ------------------------------------------------------------------
> Yes | Yes | Yes
> ------------------------------------------------------------------
> Yes | Haven't had chance to | N.A.
> | evaluate it. |
> ------------------------------------------------------------------
Notes
- (1) Should have advertised in CHEST messenger.
- (2) Not until rather late in the project. I found it by chance
whilst browsing the www.mailbase.ac.uk site. Was the announcement
made on all relevent newsgroups AND lists like sun-managers? I
only subscribe to the sun-managers list and rarely read the
computing newsgroups. Perhaps sending notification to university
computing services for broadcasting on internal university
newsgroups?
Note from Stu - We did write (both email and paper) to all UK HE
Computing Services Directors (or equivalent) asking them to
forward information about this work around their sites.
- (3) Many thanks to you for taking this responsability, I think
it is VERY important to have such a tool in the UNIX community,
because most UNIX systems do not provide such things ...
- (4) We were in contact during the evaluation period. The changes
required to NQS for our environment generally were not easy, nor
elegant, to implement although Sheffield showed willing to make
changes if necessary. The problems for us were more with the
architecture of NQS and that it scheduled jobs to queues/machines
at job submission time rather than when a queue became free. This
works okay for a large machine with several queues, but we believe
it would not work well in our distributed system where machines
generally only have one queue associated with them.
Reply from Stu - According to my records, UEA's main requirement
was cluster-wide dynamic scheduling. This feature was added
to Generic NQS 3.50.0. UEA's incorrect belief that NQS scheduled
jobs to queues/machines at job submission time shows how important
it is that documentation be produced for Generic NQS urgently.
Summation
- 27 respondants were aware of the work carried out under grant
NTI/48.2.
- 1 respondant was not aware of the work.
- 20 respondants were happy with the work carried out, although
1 respondant indicated that they were unhappy with the lack
of documentation.
- 8 respondants (all of whom do not use GNQS) did not express
an opinion.
- 19 respondants were happy or better with the support provided
for Generic NQS.
- 1 respondant was unhappy with the support provided for Generic
NQS, although they indicated that this was their problem.
- 8 respondants (all of whom do not use GNQS) did not express
an opinion.
Table 6 : The Future
> ==================================================================
> Should Maintenance + Further | Would You Contribute To Such
> Development Continue? | Work?
> ==================================================================
> Current system is probably | Probably unable to commit the time
> sufficient for our needs, but | unfortunately, unless the perceived
> maintenance would be welcome. | value of batch processing changes.
> ------------------------------------------------------------------
> Low personal priority. | Low personal priority.
> ------------------------------------------------------------------
> Of course | We already do.
> ------------------------------------------------------------------
> Yes! | Yes, with respect to porting to
> | HP-UX
> ------------------------------------------------------------------
> Yes | Not really
> ------------------------------------------------------------------
> Yes | Test future versions on several
> | platforms.
> ------------------------------------------------------------------
> Would like to learn more about| Reference site for OSF/1 (if you
> NQS | haven't got one already!)
> ------------------------------------------------------------------
> Yes, in particular ensuring it| Yes. Can act as beta-tester
> can run and be installed on a | using our machines for new
> wide platform range and | releases.
> operate reliably. |
> ------------------------------------------------------------------
> Yes. | We can consider to pay a reasonable
> | fee for the rights to use the
> | program. We can also continue to
> | beta test it.
> ------------------------------------------------------------------
> Most definitely. I would be | Yes. I do think that GNQS needs
> delighted to see Stuart | an identifiable home - whether it
> continue in his role in some | be Sheffield or somewhere else.
> manner. | That home site MUST maintain the
> | definitive version of GNQS. It
> | would be a disaster if the code
> | was to fragment into separate
> | versions. I would very much like
> | to see the mailbase NQS mailing
> | lists continue past 1997. I would
> | also likee to see the NQS WWW pages
> | maintained past 1997. I would also
> | like to see Stuart continue in
> | some capacity as GNQS developer
> | and maintainer.
> | We would be prepared to offer space
> | on our anon. ftp site and WWW pages
> | for GNQS and the time to maintain
> | them - if that would be useful.
> | We would be prepared to commit a
> | small amount of programming effort
> | but only to fix bugs that were
> | causing a problem for us or to
> | enhance the code for our own use.
> ------------------------------------------------------------------
> As NQS has fulfilled our | Our best contributions would be
> immediate needs, I personally | in the area of CAD/CAM (Catia
> do not think we have a further| AIX/HP-UX) applications with NQS.
> requirement. |
> ------------------------------------------------------------------
> Yes, *very* much | Unfortunately this is very
> | unlikely - we are currently short
> | staffed and look set to remain
> | so for some time.
> ------------------------------------------------------------------
> Yes | Beta-testing
> ------------------------------------------------------------------
> Yes definitely. My experience| I shall have to talk to my IT
> with CXbatch (based on NQS) is| manager for definitive answer
> that it does need a lot of | on this.
> support and a support which |
> does not include maintenance |
> may not be enough. |
> ------------------------------------------------------------------
> I would like to see | I would be happy to provide a
> maintenance, particularly to | small amount of technical support
> support operating system | having worked on the source code
> upgrades. The existing | in order to identify and work
> functionality is satisfactory | around an IRIX bug which caused
> so I would not be particularly| problems in the early days.
> interested in developments to |
> extend it. |
> ------------------------------------------------------------------
> Yes | No (but possibly beta-testing) (1)
> ------------------------------------------------------------------
> Yes | Yes, but only tester for new
> | versions.
> ------------------------------------------------------------------
> For those using the software | Only if we were a user.
> I would think it was essential|
> ------------------------------------------------------------------
> Yes, very important | Beta-tests
> ------------------------------------------------------------------
> Yes - at least to ensure it | No opinion given.
> will run on new OS versions. |
> ------------------------------------------------------------------
> Yes - the cost of commercial | Yes - testing and debugging on
> packages is prohibitive and | DEC-UNIX platforms.
> most of these packages are |
> based on early versions of |
> NQS. Although the code is |
> stable at present the major |
> problem is that new versions |
> of operating systems are |
> almost certainly going to |
> break the code ... without |
> support eventually all the |
> hard work will be list ... |
> Also as the nature of servers |
> change over the next couple of|
> years (i.e. parallel processes|
> become the norm) the project |
> will need to adapt to these |
> changing environments. |
> ------------------------------------------------------------------
> Yes - bug fix maintenance | No.
> would be useful. |
> ------------------------------------------------------------------
> Yes of course (for all further| Yes. I can do some testing
> development). | debugging on the platforms
> | available on our Computer Centre.
> ------------------------------------------------------------------
> Yes | Don't know
> ------------------------------------------------------------------
> Yes - we hope to make use of | We might be prepared to pay a
> it at some point. | small amount per year.
> ------------------------------------------------------------------
Notes
- (1) It should be noted that the University of Ulster has since
contributed a machine description file allowing Generic NQS 3.50.0
to be compiled on Sequent's Dynix operating system.
Summation
- 22 respondants indicated that they would like to see maintenance
for Generic NQS continue beyond the end of the JISC funding.
- 2 respondants indicated that they would not like to see maintenance
continue.
- 1 respondant did not express an opinion.
- 15 respondants were willing to contribute further to the
development of Generic NQS, mainly by acting as beta-test sites.
- 6 respondants indicated that they were unwilling or unable to
contribute.
- 4 respondants did not express an opinon.
Comparision With Previous Surveys
Introduction
In this section, we will look at the results of previous surveys [3],
[4], and the current subscriptions to the NQS-Announce mailing list
(used to announce new releases of Generic NQS) to determine how many
users of Generic NQS have not replied to this questionnaire. The
previous surveys were limited to UK HE sites only.
Previous Usage
The figures quoted here come from [4].
- Previously, 17 respondants indicated that they intended to install
and use what was then Monsanto-NQS (now Generic NQS).
- A further 7 respondants indicated that they were undecided as to
whether they would install Generic NQS or not.
- Of these 24 respondants, 10 replied to the questionnaire [2] with
which this paper is concerned. This leaves 14 respondants to [4]
which did not respond to [2].
- Of these 14 unknowns, 9 indicated in [4] that they would be using
Generic NQS, whilst 5 indicated that they were not sure.
It is therefore quite possible that an additional 9 UK sites, who
have previously indicated that they would use Generic NQS, are
continuing as NQS users.
NQS-Announce Mailing List
- On 5th June, 1996, there were 136 subscribers to the NQS-Announce
mailing list.
- This included 50 subscribers from UK HE sites (not including
the Project Officer's several subscriptions).
- Of these 50 subscribers, previous correspondance from 35 of them
is documented in either [3] or [4].
- This leaves 15 subscribers from UK HE sites, and 80 subscribers
from non UK HE sites, who have never submitted any documentary
evidence to any previous survey.
- This mailing list exists purely for announcements of new releases
of Generic NQS. It therefore follows that it is highly likely
that the majority of these subscribers are also Generic NQS users.
References
|