CCD Central Computing Division FORUM Computing Division Newsletter

Jump To Main Content

Jump Over Banner



Jump Over Left Menu

July/August 1988

RARE: The European Organisation of National Research Networks and their users

RARE is an acronym for Réseaux Associés pour la Recherche Européenne. RARE is a European organisation of national research networks and their users. Its aim is to foster cooperation between these organisations to support a harmonised communications infrastructure and so enable researchers to communicate, to use information and to access computer resources throughout Europe and in other continents.

In many European countries some form of research network service has already been introduced. These services are based on a variety of technologies, depending on local (or national) circumstances. As a result, communications with other groups both nationally and internationally can be difficult. In almost every European country programs exist to harmonize networking facilities on a national scale, but there is a growing demand amongst European scientists for cross-border communications, preferably based on the use of existing public telecommunications facilities.

Main Objectives

It is the objective of RARE to establish a communication infrastructure. This will not be achieved by building a new network with a lot of gateways to the existing ones, but by connecting and unifying all national research networks of the European countries. By its nature RARE is in a position to achieve this.

RARE supports the principles of Open Systems Interconnection (OSI), as defined by the International Standards Organisation, to guarantee interconnection of the widest possible range of systems. But the adoption of OSI standards alone is not enough to ensure communication between users. Because OSI standards are intended to be used in a wide variety of situations, they contain options and an arbitrary choice of these options would prevent interworking! One can build bridges and gateways to overcome incompatibilities, but from the user's point of view this is not an adequate solution; as more services were offered through networking, more gateways would be necessary and more inconveniences would have to be accepted.

It is in the users' interest to have facilities at their disposal that offer the same functionality for communication on a European scale as for communication within a single country.

From the user's point of view interworking of networks is a necessity. Functional standards, which specify how to make the related choices in the standards to achieve some particular style of working, are a step in the right direction, and RARE fully supports the activities within EWOS (the European Workshop for Open Systems) to establish such standards, following the work undertaken by the European standards organisations CEN/ CENELEC. However, to establish effectively a unified infrastructure for the large European research community, decisions must be taken on operational and organisational matters. It is in this area that RARE has a major role to play.

RARE Activities and Projects

  • The ongoing activities of RARE are:
    • identification of protocols and selection of implementation options
    • selection of data transmission facilities
    • directory services
    • information services
    These activities are progressed by RARE Working Groups. The Working Groups convene frequently and all members exchange ideas, papers and reports by means of electronic mail. The RARE Working Groups are active for the COSINE project in fields such as, the specification of a common set of profiles, a strategy for the management of addresses and routes, the establishment of directories and available services, the specification of testing and diagnostic tools and the evaluation of PTT services.
  • RARE organises a European Networkshop every year. The attendees are invited via the representatives of the members, to maintain a high technical level of discussion. In 1987 a Networkshop was held in Valencia, the 1988 Networkshop was held from 16 to 18 May in Les Diablerets, Switzerland.
  • In 1987 the RARE secretariat was officially opened in Amsterdam.
  • In 1987 RARE lunched two major projects:
    • the MHS Pilot Project. This is aimed at the creation of a broadly based pilot infrastructure to test the suitability of emerging electronic mail implementations of the X.400 standard
    • the COSINE project. Early in 1986 RARE was invited to draft an outline plan for the infrastructure of the European Research Network project COSINE (Cooperation for Open Systems Interconnection Networking in Europe). This plan was prepared by RARE and subsequently RARE was contracted by the COSINE Policy Group to carry out the specification phase of the project and to contribute to the preparation of an implementation plan in close contact with other parties involved, such as the CEC, PTT's, industry and standardization bodies.

Membership of RARE

There are four different types of membership defined in the RARE statutes:

  1. Full National Members of RARE are national academic and research network organisations of the 20 countries named in the statutes; Austria, Belgium Denmark, Finland, FR of Germany, France, Greece, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden. Switzerland, United Kingdom, Turkey and Yugoslavia,
  2. Associate National Members are national academic and research network organisations in other countries, which support the objectives of RARE; Hungary, republic of Korea,
  3. International Members are international organisations which support the objectives of RARE and are closely associated with the use, coordination or provision of an infrastructure for the benefit of science and research; CERN, EARN, ECFA, ECFMWF, EUUG, NORDUNET,
  4. Liaison Members are other organisations which are involved in networking and related matters and which RARE considers to be beneficial to have as members.

Apart from the above, in view of its special responsibilities for information technology and the infrastructure for research in Europe, the Commission of European Communities (CEC) is represented in RARE and participates actively in its work.

For further information see the RARE Annual Report 1987: this can be obtained from the RARE Secretariat, P.O. Box 41882. 1009 DB Amsterdam, the Netherlands, tel: +31 20 5925078.

Dr Ruud Olthoff - Publicity Officer RARE Secretariat

In general questions relating to RARE activities should be addressed by UK groups through the Joint Network Team who represent the UK in RARE. Bob Cooper (R.Cooper@UK.AC.RUTHERFORD) is a member of the RARE Executive Committee. Others should contact the RARE Secretariat in Amsterdam, + 31 20 592 5078, for information and the name of their national contact.

Using the IBM Vector Facility

The new IBM 3090-200E in the Atlas Centre has a Vector Facility (VF) which allows some programs to execute much faster by using vector machine instructions. In our current setup the VF can only be used from VM/CMS, interactively or from batch.

The standard Version 2 VS Fortran compiler (FORTVS2) can optionally generate vector code; vector versions of the NAG and HARWELL subroutine libraries have been installed, as well as a new library from IBM, the Engineering and Scientific Subroutine Library (ESSL) which has been specially written to exploit the VF.

When the FORTVS2 VECTOR option is used (with its many suboptions), vector code will be generated for DO loops if the code is suitable. Lists of programming constructs which inhibit this vectorisation and hints on removing these inhibitors are given in an IBM Manual. Future issues of FORUM will also give hints and tips. See also FIND VECTOR WRITEUP in CMS.

Although vectorisation is easy to perform, its results can be disappointing if the program is not suitable; short DO loops can take longer to execute in vector mode so care is needed. Having said that, speedups of between two and five are common, so the effort may be worthwhile.

Who should consider using vectorisation? Anyone who has a program which is written in Fortran and either:

  • Performs arithmetic on data in arrays of any rank using DO loops;
  • Uses standard numerical methods for which library routines exist (eg matrix manipulation, signal processing, interpolation, linear algebra);
  • Spends an appreciable amount of time generating random numbers, sorting or searching.

The best way to start vectorising a program is as follows.

  1. Run your program in scalar (default) mode.
  2. Run a hot spot analysis on your program to identify the routines in which most time is spent. These routines are the only ones worth vectorising. See FIND HOTSPOT WRITEUP for information on using Interactive Debug (IAD) to do this.
  3. Compile the most used routines with vectorisation switched on and examine the compiler report to see whether any vectorisation is possible.
  4. Run the program using the vector versions of your routines and compare the results and timing with the scalar version.
  5. Look at DO loops which were not vectorised and attempt to vectorise them with assistance from the manuals or PAO.

We anticipate running seminars to help explain the methods of vectorising Fortran programs.

John Gordon - User Support & Marketing, Central Computing Department

Swindon Office Profs System

PROFS (PRofessional OFfice System) has been in use at RAL since 1983. The number of users has risen steadily to over 600. It was decided at the end of 1986 that PROFS should also be implemented as a pilot project at Swindon Office; the project started officially on 1 April 1987 and the pilot project will complete at the end of July 1988. Approximately 90 staff at Swindon are now PROFS users, their names and IDs can be found by typing SHOW USERS on the command line of the PROFS main menu and then choosing the Swindon Office users option.

The Swindon Office PROFS users are using exactly the same software and computer system as the Rutherford users (namely the IBM 3090-200E computer situated in the Atlas Centre) but they access it over fast telecommunication lines with special hardware which allows terminals at Swindon to switch between the Atlas Computer Centre service and the JACS (Joint Administrative Computing Service) based in the NERC part of the Swindon Office building. The Swindon Office users are supported locally on site by the deputy project leader for the pilot project (Val Brown, SVCBO) but the software development effort and the training has been supplied by Central Computing Department staff.

PROFS is already proving valuable at Swindon for intercommunication of documents and notes, and for arranging meetings. A feasibility study has demonstrated that the Grants database from JACS can be downloaded to run under SQLDS (an IBM relational database management system) on the Atlas Centre Computer. A project is now under way to turn this into a service. Feasibility studies are just starting into the provision of information from HEIs (Higher Educational Institutions) in electronic form which would otherwise be supplied on paper forms to Swindon Office. This will not only speed the transmission of submitting information, but also ensure its speedy transmission within Swindon Office and allow information from these 'forms' to be taken off automatically into the appropriate database systems.

There is a much higher proportion of PC PROFS users at Swindon than at RAL; this has caused the project team to develop the integration of PCs with the mainframe PROFS system very rapidly.

Swindon Office management are now considering a Phase II expansion of PROFS into Swindon Office: this will (if approved) allow PROFS facilities to permeate almost the whole of Science and Engineering Divisions at Swindon in addition to the existing user community which essentially covers the higher level management at Swindon.

For further information please contact the author of this article.

Keith G Jeffery, Head Information Management Group