Produced by the Computer Board And Research Councils' Joint Network Team and Network Executive.
As mentioned in the previous issue of this publication, Networkshop 17 will be held at the University of Warwick from mid-day on Tuesday 11 April until mid-day on Thursday 13 April. The delegate list is being compiled at the moment and notices of acceptance should have been received by the time that this article appears in print. If you have made an application and not received a response by the time you read this article, please send a message to NETWORKSHOP@UK.AC.WARWICK
The Programme is very near completion and many speakers have already agreed to give presentations. The main areas of discussion are listed below, together with the specific topics that will be covered.
After the presentations on High Performance Networking there will be a panel discussion on the subject that will be chaired by one of the speakers.
On the second evening there will be three parallel discussion groups on Workstations, Network Management and Practical Mail. All of these groups will produce reports that will be presented on the last day.
As in past years suppliers and community developers will be attending Networkshop to demonstrate their products. The theme this year is High Speed Networking, by which we mean Pink Book CSMA/CD and X.25 at speeds in excess of 64K. There are plans to demonstrate the NRS campus name-server, directories, a print server and SSMP on a range of hardware including the Apple Macintosh.
Networkshop 17 is being jointly organised by John Dyer, on behalf of the Joint Network Team and Tim Clark, on behalf of Warwick University.
This article concerns the strategy for the lower layer transition to Open Systems Interconnection (OSI) within the UK academic community, and gives an update on progress in achieving the transition. A wide range of specifications, developments and service-related activities make up the overall transition, and involve all sections of the JNT and Network Executive and many members of the academic community, as well as suppliers to the community. It is not possible to go into the details of each part and hence where appropriate reference is made to other documents. Also this report should be seen as a snapshot of the work. For up-to-date information on a particular development, the appropriate section of the JNT/NE should be contacted.
In July 1987 a report entitled Transition to OSI Standards produced by the UK Academic Community's OSI Transition Group was published. It contained a strategy for the UK Academic Community's transition from existing, mainly UK-specific, Coloured Book protocols to open systems services and protocols based on ISO standards. It was recognised that the transition needed to be phased because of the different levels of maturity of the various target OSI standards. In addition the lower layer protocols up to the network level are constrained by the nature of the subnetworks involved and internetwork gateways. This internetwork capability is provided by the Yellow Book specification in the Coloured Book architecture and Yellow Book gateways are used between subnetworks. The equivalent capability is provided by OSI network layer relaying in the OSI architecture, and hence the Yellow Book specification can be regarded as a network service in the OSI context. More precisely, the type of service currently supported over JANET and campus networks, as defined in the Yellow Book, corresponds to the OSI Connection-Mode Network Service (CONS) in the OSI architecture, and it is for this reason that the basis of the lower layer transition is from Yellow Book to CONS.
The open systems networking environment that has developed in the UK academic community is complex. The major elements are shown in Figure 1.
Some 130 sites are interconnected via the wide area JANET X.25(1980) packet switching service, and there are gateways (G3) to PSS for national and international X.25 access. Most sites have one or more X.25 packet switches from multiple suppliers on their campus which share the same X.25 address space as JANET, and choose not to use Yellow Book gateways. Only a small proportion of hosts have direct connections to JANET. Connected to the LAN X.25 infrastructure are approximately 2000 terminal concentrators (PADs) and 1000 hosts (H) supporting over 3000 services. Some sites have installed campus LANs using Cambridge slotted ring technology which have Yellow Book gateways (G2) to the X.25 packet switching environment. A rapidly growing number of sites are installing a campus-wide CSMA/CD (Ethernet) infrastructure as a second generation open systems LAN technology. It was recommended that X.25-based protocols supporting CONS should be adopted for open systems communications over CSMA/CD technology and hence no transition from Yellow Book is needed for this technology. Gateways exist for the connection of CSMA/CD networks to the Yellow Book X.25 packet switching environment (Gl).
The nature of the lower layer transition can be described as the need to potentially change the LANs, WANs, attached equipment and gateways from Yellow Book to CONS whilst maintaining continuity of services. In the words of the Transition report 'the transition is feasible, but it will be a long and delicate process.....remaining with the existing interim standards would leave the community in an increasingly isolated position, while the potential long-term benefits of the move to international standards justify the effort.'
In order to realise the transition at the network level, a number of key activities were identified in the OSI strategy which support individual site transitions. Details of these activities, associated addressing issues and their status, are described in this section. Their impact on campus activities is described in the following section:
Additional features of the X.25 Recommendation introduced in the 1984 version are needed to support the OSI Connection-Mode Network Service. An OSI Requirements Note (ORN1) was produced which gives details of these requirements at the packet level. It is necessary to implement these features on the equipment used on the JANET network, and software providing this capability has been delivered and introduced into the JANET packet switching exchanges. Further information on the detail X.25(1984) and the JANET upgrade were given in an article in Network News number 27 entitled X.25 Update.
One of the factors associated with a transition from one protocol set to another is the need to maintain continuity of service between equipment using the old protocol set and equipment using the new protocol set, via the appropriate converter. When one considers the numbers of hosts and services which are involved in the transition, it is clear that some automation of the conversion is necessary. Because of the existing provision of information on available services and associated routing information in the Name Registration Scheme (NRS) central database, a mechanism exists to support this automatic conversion. Extensions were needed to describe the OSI environment and to access the information on conversion via a lookup protocol. The way in which the additional information is used, in association with the lookup protocol, is described in the Transition report (pages 18-20). The work on these extensions is being finalised by technical experts associated with the NRS project. In addition a development is in progress to provide a campus NRS server which will reduce the demands on the central servers and allow the introduction of local site specific information. An article on the development of the campus NRS server was published in Network News Number 26.
Wherever OSI networks which implement the X.25(84) protocol to support CONS must interwork with Yellow Book/X.25(80) networks, conversion facilities are required. In the first instance, gateways providing this conversion were needed when CSMA/CD networks were to be connected to the X.25(80) packet switching environments on LANs and JANET (e.g. host HI communicating via gateway Gl with host H2 in Figure 1) and gateways exist which provide this function. With the development of NRS campus servers and the further specification of the NRS Lookup Protocol (NLP) the current gateway suppliers are being encouraged to develop NLP facilities which will automate the address mechanisms at the gateway and give access to the full NRS database in the campus server.
Within the X.25 packet switching environment, it is necessary for hosts that can. support X.25(84) to be able to interwork with hosts that only support Yellow Book/X.25(80). There are several possible solutions to this requirement.
A specification of a Yellow Book/X.25(84) only converter, without full details of the NLP, was produced some time ago and put out to tender. However the relatively high development costs and specialist nature of the tendered equipment has encouraged further consideration of the other alternatives (options (1) and (3) above). Insufficient detailed information on suppliers' X.25(84) host products is available to determine the proportion of hosts that can support dual protocol operation (option (1)). It is believed that in at least two cases dual operation would require multiple X.25 front ends. It is likely therefore that some conversion will be required but the level cannot be deduced at present. Discussions are in progress with potential suppliers of general purpose and specialist gateways in the context of the updated NRS specifications.
When X.25(84) hosts in the JANET addressing domain wish to communicate with X.25 hosts outside JANET (e.g. H2 to H5 in Figure 1} the JANET/PSS gateways are involved. Currently the gateways operate at the Yellow Book/X.25(80) level on each side. It has been decided that in order to reduce the development requirements and related support costs there will only be two types of gateway, namely Yellow Book/Yellow Book and X.25(84)/X.25(84), i.e. Yellow Book/X.25(84) conversion will not be provided at the JANET/PSS gateways. Thus one of the dependencies related to the availability of the new JANET/PSS gateways is the availability of the necessary features of X.25(84) on PSS. An announcement concerning X.25(84) facilities on PSS was made in 1988, but no service date has yet been given. The Network Executive are in liaison with British Telecom on the development plans and field testing of PSS/X.25(84).
The change from Yellow Book addressing, which has been used to provide routing information on a hop-by-hop basis, to OSI global addressing has a major impact on the operation of hosts and gateways. A description of OSI network layer addressing and the reasons for the choice of the ISO DCC Network Service Access Point (NSAP) address format by the UK academic community is given in the Transition report (pages 13-16). More recently an OSI Requirements Note (ORN6) has been produced which gives further details of the OSI network address format that was chosen and outlines the address allocation scheme to be used on a per site basis.
There has been an evolution of the academic community's use of OSI network layer addressing since the original specification of OSI protocols over CSMA/CD networks some time ago. At present within a Yellow Book domain such as JANET Yellow Book addresses are used, and within an OSI domain, such as a CSMA/CD network OSI global network addresses are used. For calls from the Yellow Book domain to the OSI domain, the OSI addresses are identified by the prefix ISO. For calls from the OSI domain the abbreviated form NRS name and an indication of the higher layer protocol context are encoded within a local format OSI network address because Yellow Book addresses are in general too long to fit into the limited length NSAP addresses. With the availability of the NRS Lookup Protocol, the NRS abbreviated form can be used to obtain the Yellow Book route from the NRS Campus Server. With the recent extensions to the NRS to include OSI addressing and to the NLP specification it will also be possible to dispense with the abbreviated form NRS mechanism and adopt the same form of OSI global network address that is used within the OSI domain. This change would involve further NLP development in the gateways. These mechanisms would apply both to CSMA/CD/X.25(80) gateways and X.25(84)/Yellow Book converters. Less complex NLP requirements are involved at X.25(84)/X.25(84) gateways since the same OSI-only network address format is involved on both sides of the gateway. Details of the various addressing conversions are available from the JNT.
As well as progress on developments related to the key transitions there has been progress on a number of developments specific to key LAN components. For example:
The Transition report suggested that sites should plan to upgrade their local networks and gateways so as to be able to communicate over JANET using the Network Service support facilities of X.25(84). In view of the complexity and phasing of such changes, it may be worthwhile for a small number of pilot sites to put their plans into practice first, before there is a wholesale changeover of equipment within the community. It was recognised that one aim of the internal conversion plan for a site should be to ensure that the phasing of changes minimises the demands made on centrally provided converters. Thus, for example, where dual protocol operation can be supported on major hosts, this will reduce the amount of protocol conversion that is needed, and such dual operation should be encouraged.
Over the past couple of years there has emerged a series of activities concerning OSI functional standards, which seek to tie down the options and implementation details in several layers of ISO standards and hence maximise the chances of interworking between implementations. It is now becoming commonplace for manufacturers to make reference to one or more of these functional standards or associated specifications. Functional standards exist at the European level (they have the prefix ENV or EN) and will exist at the world ISO level (with the prefix ISP). The UK government has added to these functional standards with its GOSIP specification. None of these specifications are intended to be a complete procurement specification, and indeed the UK GOSIP specification has a separate OSI procurement handbook. In particular it is recognised that each community making a transition to OSI is likely to have requirements specific to its transition and starting environment. For example this is the case in the UK academic community vis-a-vis Coloured Books, X.29 terminal access, Yellow Book addressing and the NRS. Table 1 gives a rough guide to the correspondence between the academic community's lower layer specifications and other lower layer open systems specifications that are sometimes referred to in the UK.
| X.25(84) Packet Switching | X.25(84)LANS (> 2 Mbps) | |
|---|---|---|
| ISO Standards | ISO 8878, 8208. 7776 | ISO 8878, 8881, 8208, 8802 |
| UK Academic Community Requirements | ORN1 + PSS Compatibility | Pink Book (CSMA/CD) Peach Book (Cambridge Ring) |
| UK GOSIP Profile | GOSIP WAN (Layers 1-3) | GOSIP CO LAN (Layers 1-3) |
| European Functional Standard | ENV 41 104 Part 2 (Layers 1-3) |
ENV 41103 (CSMA/CD) ENV 41 108 (Token Ring) (Layers 1-3) |
| ISO Functional Standard | ISPTxlll (Layers 1-3) |
ISP Tx51 (CSMA/CD) ISP Tx53 (Token Ring) (Layers 1-3) |
| BT ONA profile | Network Service/PSS ONA RD004:Parts 1 & 2 |
CO LAN profile ONA RD004: Part 6 |
The lower layer transition to OSI is probably the most advanced in terms of specification and emergence of products. However it is clear from the above description that a large amount of coordination and phasing of the many changes involved at the wide area and site level is needed, in order to ensure the continuity of services and connectivity, and to limit the requirements on converters. It is hoped that the community and suppliers of equipment and services to the community will be able to cooperate so that it will be possible to continue to provide, and also expand, the range of networked services based on open systems specifications that are available to users.
Les Clyne Joint Network Team
The December 1988 meeting of the Computer Board agreed to fund the purchase of a community wide licence for the Rainbow Software which provides support for the Coloured Book protocols for use on IBM (and compatible) PCs under PC-DOS and MS-DOS. This licence makes the software available to all organisations within the CHEST remit. The Board have also agreed to fund the ongoing support of the software for at least three years. This ongoing support will provide technical assistance, software maintenance and any minor enhancements.
The overall aim of Rainbow is to provide support for the Coloured Book protocols over a range of different networks but to provide the same user interface irrespective of which network is in use. The facilities provided under the bulk deal can be split into two sections:
Rainbow provides the following:
Each of the applications are available for the network types noted below.
Each of the network support packages include facilities to enable it to be configured and incorporates comprehensive tracing and diagnostic facilities to help isolate and solve any problems which may arise.
To enable the Rainbow to be as open as possible and to be fully exploited we publish interfaces at two levels.
The interface between the applications and the services offered by each network support package is defined in the document Rainbow Network Interface Definition which is intended to provide all the information needed to add a new application or provide a new-network support package. We can also provide a two day course on using and exploiting the Rainbow Network Interface (RNI).
The LLC interface definition between Pinkbook support and an adaptor card is also available and provides the information needed to utilise almost any ethernet adaptor without modifying any of the Rainbow components.
Each of the enhancements noted below will be available at no charge under the bulk deal with the community.
Rainbow is not a public domain product, and will only be made available to a site once it has signed a site licence. This licence will allow it to be used freely within the organisation. Similarly the supplied documentation may be copied and distributed only within the organisation.
It is intended that the main distribution mechanism will be electronic. On signing a licence an organisation will be given access to the system which holds the current Rainbow software which can then be fetched using either FTP or Kermit. (A set of discs will also be available at a cost of £80 per set). A site will be required to nominate two technical contacts who would be expected to filter problems before reporting them.
One full set of documentation will be made available to each site (on paper) and it is hoped to be able to provide machine readable services on the distribution system. The current documentation set is expected to be:
In the case of an X.25 campus network a solution is to use Rainbow/LAN with a system acting as a Rainbow Server to connect the PC's on the LAN to the campus network. In the case of an ethernet based campus network however there are two different approaches which are possible:
The package which costs £190 exclusive of VAT consists of>
Type 4100-0 ISOLINK PC Controller Card, ISOLINK Binder containing:- ISOLINK Multi Protocol support software and documentation
All orders for your institution (or cluster in London) should be co-ordinated through the Computer Centre. Likewise any support queries should be through the single point of contact at the Computer Centre.
The offer consists of two options.
1. Standard Option
2. Evaluation Option
Orders should be placed through CHEST
Pat Moran Edinburgh University Computing Service
Introduction of a 3rd PSS Gateway
We are planning to bring a third gateway system into service in the immediate future. This machine will be accessible from the Manchester JANET switch. The Gateway will be accessed from PSS with the address 234223519191(69).
In parallel with this, we have decided to introduce a more general change to the service, in the hope of providing a more reliable service to users and to simplify the gateway addressing. To achieve this, we will introduce a scheme under which each user account will be available on all three systems. The service will then have only one JANET address, namely 000000000040(96), and the network will normally route each user to one particular gateway system. If, however, there is a network failure, we have the option to automatically re-route a call to another system. The only drawback in this scheme is that each call will result in an allocation change only on the gateway through which the call was routed. Under normal circumstances, a user will see no difference, but, if redirection occurs, there is a possibility of the monthly allocation being exceeded and additional call charges incurred. We plan to take steps to deal with this in the next release of the software, following representations from several users.
We will provide a mechanism for updating passwords on all systems regardless of which system the change is introduced on. During the transition stage, it is possible that some users may experience authentication problems. This is because we are unable to de-encrypt the password files, and users who get routed under the new scheme to a different gateway to the one they normally use, may not have the correct password. If you experience authentication problems then we suggest that you call your normal gateway, modify your password there, and wait a short time for the change to propagate before attempting to use 00000000004096. If you still experience problems, please contact Bill Jenkins or Shirley Wood, of the Network Executive or send email to W.Jenkins@UK.AC.RUTHERFORD or S.Wood@UK.AC.RUTHERFORD.
Ian Smith Network Executive
I have been asked by the ULCC NOC to remind users of the London PSS gateway that the PSS address in use for the gateway is 234219200100. The old number which used to re-direct to the above number is now in use for a different purpose and may give varying responses if called. In future incoming calls from PSS to the London PSS Gateway will be rerouted to either the Rutherford or Manchester PSS Gateways, if problems are experienced on the London PSS line.
John Carey Network Executive
In Network News No. 27, Bob Cooper announced a revised authorisation scheme for electronic mail. This has now been implemented, and so it is no longer necessary for a user of a mail system within the UK academic community to register with the gateway (provided his mail system is registered in the NRS and appropriate funding to support his activities is available.)
The new link between ULCC and the USA should be installed in the near future, and we plan at that time to rationalise the naming of the service. This has not been done earlier because the US side of the connection still comes via UCL. The new service name will be UK.AC.NSFNET-RELAY, but the old name of UK.AC.UCL.CS.NSS will be supported for a month after the new name is brought into use, in order to allow a smooth hand-over. Those of you whose Mailers make automatic use of NRS Partial Domain information (i.e. who do not have to explicitly indicate the route through the gateway in the To: field, or any other recipient field of the mail message) should not have to take any action as appropriate changes in the NRS will take care of everything for you. Those who must specify the gateway name as part of the To: field will have to change once the new link is brought into service, but please don't try until then. Users in the US in general have domain-based mailers and so in the majority of cases the network providers will be the only ones who need to make changes. We encourage all of you who currently send mail to Internet sites via EARN to switch to using this route, which we believe will be a quicker and more reliable link to the Internet sites.
Once the new link is installed, it should be possible to allow FTP and TELNET access to those who would like it. However, we need to point out that this can currently only be achieved by a staging process and it is necessary to have an account on the gateway. Accordingly, we ask users to send files using electronic mail wherever this is possible (especially for text) and to only ask for file and terminal access where it is absolutely necessary.
We plan to tidy up the registration data base, and remove individual accounts wherever possible i.e. for those who only need MAIL. We would be grateful for your assistance in telling liaison@.UK.AC.UCL.CS.NSS that you no longer need the account.
Ian Smith Network Executive
The ISO Virtual Terminal (VT) Standards are an attempt to decouple the particular design of the terminal from the datastream that flows between it and the host. In this way, many of the problems associated with the use of proprietary terminals can be avoided.
VT Standards have received little publicity within our community, other than the length of time it has taken to reach agreement on a stable standard. This article is intended to bring you up to date on the Standards and give some indications of how they might be used within the UK Academic Community. It is not intended to provide tutorial information.
There are two VT standards:
The term Class refers to the type of data the terminal is capable of displaying and manipulating. When the standards work was begun in the 1970s, it was envisaged that there would be many classes, one each for Basic Class, Forms Class, Graphics Class, Text Class (in ODA structure) and Image Class (bit mapped). It became apparent however that it was an impossibly large task to define all of these classes at once and standardisation began on just one - Basic Class.
Basic Class defines a way of dealing with a rectangular array of character boxes and their contents. The original Basic Class in itself contained so little functionality that it appeared to be of little use. Proposals were made by bodies interested in form filling operations that features appropriate to Forms Class be added to the Basic Class to make it more useful. This work was undertaken as an Addenda to the main standards. The Addenda provide extended facilities, principally the ability to define Blocks, Fields, and ways of controlling data entry into the fields.
The final text of both 9040, 9041 and their Addenda has now been agreed. It was decided sometime ago that the text of the Standards and the Addenda should be merged before they are published in final form. This merge is at present under way and a final fully merged text should exist around the middle of 1989. ISO will probably publish the official typeset IS documents around the end of the year.
It also became clear that the functionality provided even by the merged standards would not provide the type of facilities required by academic community users undertaking full screen text editing. A second Addenda often referred to as the Ripple Addenda, which amongst other things will support Ripple Mode editing, including cut and paste operations has been proposed. The status of this work is that there is. currently a working draft which will probably reach DP by the Autumn of 1989.
No formal work has been undertaken on any of the other classes, but I have heard there is interest in pursuing the Graphics mode in the United States.
Some aspects of Virtual Terminal characteristics are not fixed by the Base Standards, for instance the structure of Data Stores and Control Objects. Instead a number of parameters are defined, which when assigned values, define what is referred to as a Virtual Terminal Environment (VTE). For the sake of efficiency and as an aid to interworking., profiles are being agreed that define complete sets of VTE-parameter values, some of which may optionally be supplied as arguments. Two of the regional workshops on Open Systems have been working towards a series of VT Profiles. In the United States, The National Institute for Standards and Technology (NIST) has been working on several profiles, the most important of which for European interworking is Forms Mode. The European Workshop on OSI Systems (EWOS), is currently working on two profiles, Page Mode and Forms Mode, with plans to develop enhanced versions of each to include the Ripple extensions in the near future. There is strong liaison between both NIST and EWOS and a commitment on both sides to align the profiles. UK GOSIP has decided that rather than define its own VT specification it will now provide a pointer to the regional workshop documents.
There are two modes of operation defined in the Standards, Asynchronous (A-mode) and Synchronous (S-mode). A-mode has gained favour in the US, and represents character by character interworking. Two display objects are required and the synchronisation of data between them becomes a formidable problem. In Europe S-mode will be adopted, indeed the EWOS Expert Group on VT are only working on S-mode profiles. S-mode operation involves block mode transmissions with only one display object which can be updated in a two way alternate fashion by the holder of an access token.
VT Standards do not however define the ways in which a full screen service can be offered in a windowing environment. This is being looked at in ISO by the Terminal Management Group. Terminal Management (TM) work has been going on for some time now at both the national and international level. TM will provide tools for the management and coordination of a number of associations between a terminal and host(s). The associations may carry data of the same or different types (e.g. text, graphics, VT. sound). It is thought that the first IS text in The TM area will be published in 1991, however it is unlikely that it will be complete in all parts until the year after.
There has been some interest in X/Windows both within our community and the UK standards organisation. The JNT has commissioned a feasibility study into the possibilities of running X/Windows over CONS. It would seem that this is entirely possible over high, bandwidth local area networks, but suffers from severe traffic problems over slower technologies. The standards bodies also recognise the latent need for an open and standard way of providing a windowing environment right now and are actively looking at what might be done to satisfy this demand quickly. Watch this space !
The JNT has been active in the Full Screen Services field for some while. The Simple Screen Management Protocol (SSMP), published in 1985, now has a large number of implementations which bear out its usefulness and efficiency. As an interim I would encourage further implementations, particularly where a general purpose driver can be inserted into the operating system to provide a full SSMP service of all existing facilities, such as has been done on VM/CMS and is being undertaken for VME. It would be useful to push this work further and include other major community operating systems such as VMS, Primes and Unix.
As a first step towards providing a general VT service, the JNT are planning to commission a VT pilot project. It is likely that it will centre around portable implementations of a VT terminal and an SSMP to VT gateway, both based on MS-DOS microcomputers.
In the European forum we are contributing to the RARE Working Group on Full Screen Services and will be taking part in their pilot study which is currently being discussed. The other countries that are actively participating in the VT field include West Germany, The Netherlands, Spain and Switzerland.
If you are interested in finding out more about Virtual terminals I can recommend an introductory guide that has been published by the DTI, copies of which are available from me at the JNT.
I intend to start a VT-Implementors email discussion list in the near future, those interested should forward their email addresses to VT-Implementors-Request@UK.AC.RUTHERFORD
John Dyer, Joint Network Team
Glasgow University Library is pleased to announce that it has been awarded the contract for Project Jupiter, a UGC-funded project sponsored by the JANET User Group for Libraries (JUGL). The project is designed to increase the knowledge and use of the JANET network among academic libraries.
John MacColl will be seconded from the Library staff to the project for two years. He will be supported by colleagues both, from the Library and from the University's Computing Service Department. John graduated from St. Andrews University in 1983, and obtained a Diploma in Librarians hip from Robert Gordon's Institute of Technology, Aberdeen, the following year. He has been on the staff of the Library since October 1984, and has worked in both Technical Services and Reader Services. He has gained wide experience of online searching and been closely involved with many aspects of library technology and staff training.
The Project objectives are:
There will be six regional training seminars - two each in London and Birmingham, one each in Manchester and Glasgow. Each seminar will cover hardware requirements. communications software, local, national and international networks, present applications and integration of JANET into present library practices. Seminar locations will have facilities for adequate hands-on experience.
Areas of further research and development will include the use of JANET for file transfer of catalogue records, access to Online Public Access Catalogues (OPACs) abroad, further streamlining of inter-library loan operations and the use of JANET for document delivery. Liaison with organisations such as the British Library, the Consortium of University and Research Libraries (CURL), and the Standing Conference of National and University Libraries (SCONUL) will be sought.
The Project will also endeavour to set up formal links with Library User Groups of international networks with the aim of fostering similar services on less developed systems.
It is hoped that proposals will be brought forward for the development of a continuing programme. The intention is to provide a forum for ongoing training to disseminate information on developments within JANET and related areas, and to act as a pressure group and think-tank for the development of new initiatives. For further information please contact:
John MacColl, Glasgow University
It has been pointed out that as lines are upgraded to higher speeds, the old modems become available for re-use elsewhere. In some cases it may be possible that they will become available for loan to university and other academic sites. Any site wishing to give a home to a modem or modems should contact their local NOC who may be able to help them.
John Carey
Network Executive
Thank you to Hewlett Packard. Apollo (FTM Communications) and Torch for their updates to the JNT protocol information. Unfortunately the response from suppliers was not enough to make it worthwhile producing the table in this edition of Network News.
Information on the updates of the three suppliers named can be obtained from JNT-A@UK.AC.RUTHERFORD on request.
Collette Truman, Joint Network Team)
A survey of Ethernet repeaters conducted, on behalf of the JNT, by Edinburgh University Computing Service identified BICC Data Networks, Cabletron Systems, and Pirelli-Focom as manufacturers of dual-port, fibre, and multi-port repeaters, all of which have now been recommended for use within the academic community. Copies of the survey may be obtained from the JNT Administrative officer (M.Thompson@UK.AC.RUTHERFORD).
The survey was conducted by examination of manufacturer literature, backed up, where necessary, by telephone calls to sales staff. None of the equipment was physically examined or tested. Purchasers should therefore ensure they have commitments from their supplier on the proper functioning and conformance to standards of their equipment.
Dick Gillman Joint Network Team
A print serving box is being developed by the University of Aberdeen. It is based on hardware similar to that used for the Fawn Boxes. The unit is self contained, with dual ports. The intention is that the server will sit on a synchronous X.25 link, running at speeds up to 9.6kbs, and will provide an asynchronous link via its second port to which a printer or similar device may be connected. Data files will be submitted to the server using either the FTP or JTMP protocols.
A bank of 16 dip switches are used to configure the device to select printer type, communication interface options, banner production, etc. Binary transfers with 8 bit word length are supported, allowing flexible output to laser printers, plotters, etc.
Anyone interested in further details should mail G.Robertson@UK.AC.ABERDEEN.TEST
A prototype box is presently being tested against a number of different FTP and JTMP implementations.
G.Robertson University of Aberdeen
There is now a new edition of the NRS Technical Administrators Guide, which all Technical Administrators should have received through the post at the end of January. The document has been extensively re-written since the 1983 edition to include new material and make the guide more understandable, especially for those new to the NRS. If you are an NRS Technical Administrator who has not yet received a copy of the new guide please send an email message to the address below.
I have already received a few comments about the guide but to aid in the planning of future revisions of this guide and a companion Implementors Guide I would find it helpful to have detailed comments from many more of you. In particular I would like to know what you think about its technical level, content and the presentation style.
Comments should be sent by email to the address below.
JNT-NRS-COORDINATOR@UK.AC.RUTHERFORD
John Dyer, Joint Network Team
Instances have occurred recently where a lot of time has been wasted at a JANET Network Operations Centre (NOC) in investigating why mail was failing to get to a particular site. It turned out eventually that this site was barring all calls made from the source address as the site had had trouble with hackers from that address. The NOCs would appreciate being informed of sites that undertake this practice. It would be sufficient to inform only the NOC to which the site is connected.
John Carey Network Executive
Those of you who, like me, are a bit long in the tooth will remember the glories of Metronet which used manufacturers job entry protocols to pass jobs across concatenated machines. EARN works in the same way but with the advantage that only the IBM Network Job Entry (NJE) protocol is used. Although the protocol is not sophisticated its implementations are highly reliable, available on a wide range of machines, and well maintained by the manufacturers.
It has proved that it is possible to build networks with thousands of machines which work remarkably well. In the case of EARN, various services have been added to provide mail and database access facilities. The mail service has become particularly sophisticated with the advent of LISTSERV which allows linked distribution lists to be maintained thus reducing network traffic.
There are many good reasons for wanting to use better protocols. Store and forward networks are no good for interactive traffic, and they are dependent on large computers. There are also political pressures from CEPT, RARE, and others dedicated to the use of a common set of protocols for the community. The delay has been due to the lack of products and finding ways of insulating users from changes.
The low level protocol to be used is not surprisingly X.25. Discussions suggested that a strategy based on developments of TCP/IP towards TP4 had many unknowns and was at variance with current European ideas for wide area networks. The TP4 ideas are probably more appropriate for high speed networks (greater than 64K) and such lines look to be very expensive in Europe. X.25 is satisfactory for slow and the sometimes unreliable lines currently available.
X.25 technology is now very mature and its characteristics well known. Equipment and software is reliable and easy to obtain.
Examination of the public networks revealed that they would be impossibly expensive for EARN traffic patterns. Costs would have risen by a factor of 10. It also seems that the data rates to be expected on public networks can be fairly low. Thus a private network is being developed.
The strategy is to provide an international transit network. Each country would have its own network(s) which may be the current EARN service, a national X.25 network, or a network of some other technology. Thus there will be gateways or relays between the international part of EARN and national parts.
As an aside it is believed that this is a better model than the PTT one where each country connects to many others via bilateral agreements. With an international transit network a national network only has to negotiate with one other network and it is the international one which guarantees the bandwidth required with any other network.
Whilst the strategy looks nice and clean the practice is not so. There are a large variety of gateways and relays needed. For example:
The requirements for the switches and topology is to minimise costs and maximise performance. Unlike JANET, the line costs are very high, the number of connections low, and the cost of switches zero as they have been generously donated by Northern Telecom,
To maximise performance there should be one switch with a high speed connection to each country. To minimise costs there should be a number of switches located in low tariff countries.
The result of these conflicting requirements is to have 4 switches located in the UK, the Netherlands, France, and Switzerland. Although Switzerland doe not have low tariffs the CERN laboratory, which is a major source of traffic, is situated there. These switches are to be interconnected by a square of 64K lines thus giving a measure of redundancy. The other countries will connect to the switches with existing 9.6K lines or new 64K lines depending on demand and finance.
The switches are being supplied free by Northern Telecom - model DPN100/20S.
These are near the bottom of the range of NT switch and have the following specification:
Further developments will allow the use of 2M connections.
Variations on the above configuration are possible to meet local requirements. Other models of the switch can be very large but they are all are made from the same basic components.
The switches are managed from a DEC VAX. Operator consoles connected to a switch are only able to interrogate its operation and cannot control it. All control comes from the VAX and all alarms are sent to it. In the event of failure the management can be taken over by another VAX in the network.
A lesson quickly learnt is that switches from different manufacturers are very different and reflect the market of the supplier. In the case of Northern Telecom the switches are designed with the needs of PTTs in mind where high reliability and very strict adherence to the CCITT recommendations is demanded. The addressing mechanisms reflect the needs of the public networks in that X.I5 is provided and a network is to a great extent a set of switches with a common DNIC and X.75 connections to networks with different DNICs.
As EARN expects that eventually the international infrastructure will only interconnect other networks it is not expected that there will be a need for many ports. In the short term provision has to be made for a small number of connections into a country for an X.75, and other gateway connections.
The structure of a Data Terminal Equipment number on the international network is:
DNIC 4 digits 3 digit country code based on X.I21 5 digit address 2 digit sub address
The DNIC is 2000 but is not registered with CCITT. It is unclear whether such a registration application would be successful.
The 3 digit country codes of each country connected to the switch is allocated to that switch and any equipment in that country has that code. The country code 200 is reserved for the network management equipment which needs a set of addresses.
The 5 digit addresses are allocated from 1 upwards. The 2 digit sub address is not policed.
This structure reflects the transit nature of the network. Eventually it is expected that each country will either have an X.75 connection or a connection to a single gateway. In the short term there may be all sorts of connections to get over any temporary problems.
EARN will use the ISO DCC scheme for NSAP addresses. Where possible existing registrations of the academic communities will be used. This raises problems as there are few such registrations and it would be counter productive for EARN to make registrations as these would eventually cause some entities to have dual registrations. Thus in the short term EARN will invent interim registrations.
There is also a problem with the part of the NSAP below the organisation as there are few schemes in existence and the directory mechanisms for NSAP to X.25 DTE are not available. EARN's interim strategy will be to place the X.25 DTE in this part of the address pending further developments. This will allow an algorithmic conversion from the NSAP to the X.25 DTE.
The principal requirement is to maintain the current services. This means devising a scheme for carrying NJE across X.25.
It was not feasible to convert NJE to FTAM, or similar schemes, since these would need a lot of implementation effort and there may well be loss of quality.
The first scheme studied was to run NJE across an IBM SNA session across X.25. This would be easy as the products exist. The drawback is that the software only exists for IBM machines and about half the machines are VAXes.
The second scheme, which has been adopted, is to implement NJE over ISO session layer. This can be easily implemented on VAX and IBM computers. It has the psychological advantage that it uses more ISO protocol than SNA session.
For an IBM computer the required code has been developed by IBM in Heidelberg. This uses the IBM ISO products OTSS and OSNS. The system is now available. This is known as an E-box.
For a VAX computer under VMS the required code has been developed by Joiners Associates using as a basis the JNET NJE code. It uses the DEC OSI products OSAK. VOTS and PSI. The system is now available. This is known as a G-box.
The scheme works by the E-box or G-box operating as relays between NJE over X.25 and NJE over bi-synch. The E or G-boxes set up virtual calls between each other over which NJE operates. Unfortunately it is not easy to detect the end of an NJE document and so the calls are left up permanently. This limits the number of E and G-boxes which can be interconnected before you run out of call capacity. It is hoped to overcome this limitation with further developments which will drop connection on a time out or if the end of an NJE file can be detected.
The expectation is that there will be an E-box. a G-box or both in each country. DEC is supplying the G-boxes.
Other ISO protocols are expected to operate over the network. So far the only one sufficiently developed from the point of view of available products is X.400. It is hoped that this will operate in late 1989. A gateway between X.400 and the currently used RFC822 mail is being developed by Joiners Associates and will reside in the G-box. Details of the use of X.400 have still to be worked out but will align with the RARE work.
Although not strictly ISO some X.29 interactive use is expected.
FT AM should be available in a year or so but it is too early to lay plans. JTM and VT are not expected in the foreseeable future.
EARN believes that Europe should be provided for by a number of national and international networks.
These will provide services to all or parts of the community depending on who runs them. They should be interconnected under suitable arrangements and this implies that they use harmonised standards to allow the interconnection. The work of the functional standards groups, RARE, and COSINE will be followed if at all possible as EARN is not in the standards business.
Thus EARN expects to connect to national parts of EARN and other international and national networks. The exact nature of these connections is unclear. However, EARN would prefer X.75 connections as this follows the X.25 model which is well established. The connections will be very dependent on the plans within a country and it is not EARNs intention to influence a country in any particular direction as long as the service to the user can be maintained. None the less there are benefits in European countries following a common path.
Paul Bryant
Central Computing Department Rutherford Appleton Laboratory
The Networkshop was organised by the JNT with the much appreciated help from the Computer Unit at City University. The aim was to encourage a greater flow of information about Networking (particularly OSI) between the academic Unix community and the JNT. The workshop was attended by over 90 delegates from over 60 different institutions and a number of manufacturers were represented.
Unix has a large and increasing market share in the academic community and in many situations Unix systems offer very good value for money. The many suppliers offer a large choice of systems, which is advantageous to the purchaser, however there is not a standard Unix which often means that each machine range will require different implementations of the same software. The Academic Community and the JNT will watch the development of POSIX and other initiatives on Unix standardisation.
Andrew Dand (Bath University) gave a talk on File Access. The requirements for file access are to access shared data and provide a managed file store. FT AM will provide file access, however it is not straight forward to implement a file server using it, but developments are underway to rectify this problem. File transfer is the simplest application of FT AM and very similar to Blue Book NIFTP The initiator and responders are both stand alone processes which are not integrated into the operating system. For file access the initiator is integrated into the operating system and for file serving both the initiator and responder are totally integrated into the operating system,
Les Clyne (JNT) gave a talk on LAN/WAN issues, giving the history behind the development of the Academic Community X.25 packet switching network (JANET) and the Academic Community's strategy for Lower Layer transition to OSI. A discussion group was held to discuss this topic and the following issues were raised:
Jim Craigie (JNT) began his talk on Message Handling Systems by giving a brief introduction to the JNT/NE and Coloured Book protocols. Jim explained the approach being adopted in the transition to OSI in general and the transition from Grey Book mail to X.400(88)/ISO 10021 (MOTIS) in particular. This transition target rather than X.400(84) had been chosen because of the significant benefits involved including:
Jim then outlined the proposed replacements for the Coloured Book protocols YBTS - > OSI Connection Mode Network Service (CONS); Grey Book JNT Mail Protocol -> ISO 10021 (MOTIS).
Paul Barker (UCL) spoke on the UCL Directory Services. Directories are a database-Uke (ISO Open Standards) data communications technique for distributing information and significantly assisting information management by distributing that management to those responsible for the information. Paul then went on to give examples of suitable applications and discuss the theoretical scope of directory services.
Piete Brooks (Cambridge University) - Piete outlined the benefits of using publicly available software and how to find out if what you require is readily available from another source. Piete explained the usefulness in developing new software to be able to call on the assistance of others and subsequently making the software available to others, thus helping to prevent people wasting valuable time and effort when the software is already developed and tested.
Julian Onions (Nottingham University) spoke about the ISO Development Environment (ISODE) briefly explaining what it is and outlining a number of applications within ISODE. FTAM is available and comprises a library of file transfer and management procedures and routines. Directory Service was included as a beta release in version 4 of ISODE. Virtual Terminal Protocol (VTP) is written but is not yet part of release 5. Message Handling Services (MHS) is expected to be included in version 6. Other facilities include IMISC. a collection of miscellaneous services.
David Mercer (Edinburgh University) spoke about the principle features and problems encountered in porting and supporting Unix networking software. A brief introduction to the development of Unix and the communications used at the university and EUCS was given outlining the problems involved such as the differing requirements of suppliers, and timescales.
Collette Truman, Joint Network Team
In December a new version of the NISS catalogue was released, which featured a number of enhancements to both the user interface and functionality of the system. Although a number of improvements are still planned for the catalogue vehicle, effort is now being focused on the task of developing a useful database.
Since Christmas, significant advances in the catalogue's coverage have been made; more micro software has been added, a start has been made on cataloguing the holdings of two representative Computer Centres, and some sample records of the holdings of the Oxford Text Archive have been added.
For more information on NISS Catalogue activity mail Colin Work at NISS@UK.AC.$OUTHAMPTON.IBM
In November we began a major publicity drive, sending a mailshot to Computer Centres, Supplies Offices, Libraries and the Head of every Department in every Polytechnic. University and College of Higher Education in the UK. Each recipient was sent a CHEST Directory, information about the NISS Bulletin Board (NISSBB) and a short questionnaire. The questionnaire was devised to
Responses received to date indicate that a majority of computer users in Higher Education still know little about JANET and its services, or are unable to gain easy access to them.
If you have received a questionnaire and not yet returned it please do so; we need all the feedback we can get if we are to keep information services such as the NISSBB on the right track.
NISSBB usage has increased relentlessly since its launch in September; we are currently recording around 200 NISSBB sessions every working day. The huge November mailshot has undoubtedly helped to spread the word.
In our December software release we introduced the POST facility which makes it easy to download information from the NISSBB to your own computer system, as long as you have a JANET e-mail address. This is needed because POST sends the information to you overnight using e-mail. (If you try POST and the information doesn't reach you by the following morning it may be because you specified your electronic mail address incorrectly; if in doubt consult your local Computer Service for advice.)
Finally, a word of thanks to those Computer Services who have added the name NISS to their local PAD units; this makes it very much easier for your users to access NISSBB over JANET using an easy to remember name rather than an X.25 address. Here's hoping that other Computer Services will follow suit.
Meanwhile, if CALL NISS does not work from your PAD. you can still get to us by using the JANET address: i.e. 000062200000.
Rob Armstrong, Information Services Manager, NISS
On the 30th September 1988 Camtec Electronics Ltd was acquired by Hunting Associated Industries PLC, (HAI). Vie Toy and David Drury (Camtec Directors) along with their management team began a search for a partner in April, after concluding that with the growing strategic importance of communications to organisations many of the larger markets were closed to small private companies. The acquisition provides Camtec with a significant number of new opportunities. The financial support and backing of a major PLC will allow the company to plan with confidence for the long term future to the benefit of both the company and the customers it serves.
Camtec with its extensive communications skills, joins Lynwood, a major supplier of terminals and Unix workstations which was acquired by Hunting in June 1988. The two companies already serve many common customers for example in Government and the Police forces. With the combined strengths of Camtec and Lynwood and with other acquisitions the group as a whole will be able to provide effective business solutions to a range of markets in the 90s.
Hunting Associated Industries PLC with Hunting Petroleum Services PLC and Hunting Gibson PLC form the Hunting Organisation with total annual sales exceeding 350M. HAI is an engineering company with four main areas of activity, aviation (the Field Aviation group of companies) defence (Hunting Engineering), composites (Halmatic) and most recently Information Technology (Hunting Electronics Ltd).
Keith Rushton, Managing Director Hunting Electronics, sees his mission as building by acquisition and organic growth a significant force in the European Information Technology market, with the full support of the Hunting organisation. With Lynwoods' expertise in information presentation now coupled to Camtecs' expertise in information transportation a good start has already been made.
Vic Toy and David Drury remain on the Board of Camtec and look forward to playing a very active role in the development of Camtec and Hunting Electronics over the next few years. They are joined on Camtec's Board by Keith Rushton as Chairman and Steve Hibberd (Hunting Electronics Company Secretary). Richard Haunch our Finance Manager becomes Company Secretary. The final member of the new Board will be the General Manager, a position currently undertaken by Phil Corbishley, a consultant.
Camtec believe they were fortunate in finding in Hunting an organisation whose style is to provide financial and managerial support to independent managements enabling them to fulfill their own plans. Within Camtec the implementation of policy set by the Board will continue to be delegated to the General Manager and his management team, namely John Donaldson - Sales and Marketing, John Pragnell - Engineering, Bill McCrorie - Support and Richard Haunch - Finance.
The practical benefits of the acquisition are already beginning to show internally, with both Sales and Engineering teams being significantly strengthened, with a substantial investment programme in people and equipment being put in place.
Camtec plans to build on its strengths in X.25 and Ethernet products and systems and continues to see support for ISO standards as a cornerstone of company development policy. The Sales and Marketing team look forward to being able to discuss with all our existing customers how we can best meet their future requirements not only in the short term but in the longer term through the 90s. Engineering are already actively looking at the next generation of technology for WAN's and LAN's and will be looking for collaborations for the development of advanced products for the 90s.
David Drury Camtec.
