Open-system standards progress The second meeting of the ISO Committee concerned with the area of open-system interconnection (TC97/ SC16) took place in London during June 1979. Considerable progress has been made since the first meeting in Washington in February-March 1978 =. There, the urgency of such standards was pointed out, as was the need to prospectively define such standards before proprietary standards proliferate. As the bounds of possible computer systems increase rapidly, and new areas such as word processing, and facsimile transmission emerge, and the interworking of such equipment with data processing and technical computing equipment becomes at least desirable if not necessary, open-system standards are becoming more and more important. 2 Such standards are seen as the key to a number of future applications of computer systems that will involve the unrestricted or minimally restricted interconnection of large numbers of subsystems distributed internationally across a wide range of fields of application ranging from electronic funds transfer, through electronic mail and office automation, to particular applications-oriented systems such as airline management systems. The UK view reflected in the previous article ~ is that the upper and lower levels of protocol should be decoupled, so as to allow developments to occur above and below such an interface independently. Above the interface it is felt that separate protocols for file transfer, terminal handling, session establishment, and other 'high-level' functions should be developed, to be used as needed. The basic needs governing the development of such protocols are transparency, i.e. a facility that allows their use without knowledge of their working and detailed design2, freedom from errors, the ability to link various services (i.e. network independence) and the ability to satisfy a wide range of requirements. During its second meeting, the attention of the SC 16 Committee was concentrated on refining the
244
architectural model presented in differing forms at its earlier meeting; on unresolved issuesoutlined in more detail below; and preparation for direct work on the required standards within short time scales, if or when SC 16, and/or its other counterparts within the ISO are authorized to carry out such work.
Contributions to the London meeting, There has been a considerable level of activity within a number of participating countries since the previous meeting. The following notes reflect the position of various participating countries as of June 1979, although it should be noted that some of them were not present at this meeting, and therefore their written contributions have not been included in this short report as they were not presented personally (in such a way that the meeting could interact with the presenter). The following reports were presented at an early stage in the meeting.
Canada Canada has concentrated on issues remaining within the model for interconnection, including the interconnection of networks, the position of X.21 and X.25 within the architecture, the viability of sublayering, and the boundaries between presentation and session layers. These boundaries were very considerably tidied up during a later part of the meeting.
Denmark The main concern expressed by the Danish delegate was that there are areas where no specific requirements or guidelines have been identified. This remark was based on experience, as one large Danish service company has attempted to specify a large network based on the previous sevenlayer model. In view of his country's
experience, the Danish delegate pointed out that any approach is allowed, in terms of implementation, where no guidance is given.
France The French delegation wished to see the reference model document processed as, and issued as, an international standard. This proposal was considered later, but did not receive unilateral support.
West Germany Germany felt that there was a need for a basic reference model (simpler than the seven-layer model) to be prepared urgently. Further, it felt such a model should be easy to understand, and clearly identify areas where there is such a need. Such a model might be sublayered at a future date. A view is also expressed that the present model document is large and unwieldy - and needs considerable reduction and simplification before it will be acceptable to those who will need to use it. Italy The Italian delegation opposed the processing of the model document as a standard, and went on to express concern that the model document should take into account the transition to open-systems architecture from the existing status quo. They reiterated the point made earlier that the model must necessarily cover facsimile, text transfer, and word processing.
Japan The Japanese delegation felt that the reference model must be completed with urgency, and two of the higher priority tasks are the definition of layers and sublayers, and the position of applications within the whole structure. It also highlighted the need to start work now on standardizing protocols that will later fit into the completed models, and also the importance of liaison with other interested groups such as TC97/SC6 and the CCITT Study Group VII.
0141-3664/79/050244-05502.00 © 1979 IPC Business Press
computer communications
The Netherlands
LIAISON BODY REPORTS
The reference model was stated to be highly complex, and in need of simplification.
IFIP
Norway Although not a participating member of the Committee, the Norwegian delegate was asked to give his views. He again felt the reference model should be simplified and reduced in size, and be clarified.
Sweden Three aspects of standardization were identified: the quality of a standard, its acceptance, and the timing of its availability. Sweden believed that there is too much accent on time in this case, and quality may yet be lacking. It pointed out, correctly, that SC16 has still no authority to develop the standards, and should concentrate on its present study phase of work.
UK In appropriate contributions, the UK highlighted the present inadequacies of the model, indicating that is is somewhat rigid. Other separate contributions highlighted the need to identify the position of encryption within the architecture, and the need to review the basic model structure.
USA The USA foresaw a need for a set of standards to be at once flexible, allow for growth, and not obsolete previous work. It felt that the state of the present model document, considered unsatisfactory, was a consequence of hurried editing. Major issues often have to be resolved at ISO meetings by ad-hoc groups during the last day or so of a meeting, and this procedure does not always produce the best results. It felt that SC16 needed more active liaison with other groups involved with the same area of work such as various TC97 sub-committees, TC95 and the CCITT.
vol 2 no 5 october 1979
The I FIP spent a considerable amount of time trying to rationalize the model, and at one stage considered reducing the number of layers by combining present layers 3 and 4, and 5 and 6. Having very carefully investigated the consequences of such actions, they have rejected the idea, stating that they presently favour the retention of all seven layers. Further, the IFIP favours the inclusion of multi-point and broadcast connections - a point for further study. It further supports the initiation of work on specific areas such as virtual terminal protocols at an early date, as these are urgently required.
ISO Committee TC97/SC6 The representative mentioned two projects of the committee that are more closely related to this area Projects 17 and 24. Project 17 has led to production of a guidelines for layout of header format, on which considerable further liaison will need to occur. Project 24 has led to more detailed recommendations for communications format headers intended for levels 3 and 4, although it is felt that a family of heading standards might be developed for a more widespread use. The SC6 committee believes that a top-down approach is necessary, though the model must be general and open, and also belives that SC16's main task is the understanding and extension of the model, and generations of new items of work that need not necessarily be progressed by that committee.
ECMA The European Computer Manufacturers' Association outlined their contribution, stating that they now think it is time to start work on standards. They believe that the present model should be accepted as a framework for most of the urgent standardization work. It is particularly important to avoid the proliferation of standards - an interesting viewpoint coming from a manufacturer's
group! ECMA already has task forces working on virtual terminal protocols and virtual file protocols, and is undertaking some work in the session layer areas. In particular, they also wish to take into account the services provided by CCITT. CCITT The representative of this body stated that they have already examined the application of the model to facsimile transmission and word processing. CCITT urgently needs a structural model, and hopes to decide on an appropriate structure by the end of 1979.
ISO T C 4 6 The technical committee of the ISO concerned with documentation is already examining the interchange of information via magnetic tapes, basic character sets, and the number of escape sequencesassociated with basic character sets. It is therefore working on a number of standards that should fit (and indeed will need to fit) into the open-systems reference model.
REFINEMENT OF THE MODEL The major part of the meeting was devoted to refining the present model, in the hope that it might be finalized by the end of the meeting. In the event, this proved not to be possible, owing to the number and complexity of some of the modifications required. A considerable amount of clarification of the model document occurred, including the addition of text to cover aspects such as 'binding' (provision for two or more applications to interwork); it is now recommended that this be arranged wholly within the application layer. An important extension of the model was addition of a new Annexe, summarized schematically in Figure 1, indicating how the various elements involved in open-wstems interconnection overlap at the applications layer and then use common services below that layer. Certain aspects raised within the additional text involve further study, such as the use of
245
I \ \
Program support
// ation ~\
J
)/
layer
Presentation layer Session layer Transport layer Network layer Datalink layer Physical layer
Figure 1. Schematic diagram of the relationship between entities at the upper level and services below (the overlap of the entities is asymmetrical, and presently not defined as a detailed level) nontelecommunication transfer mechanisms i.e. offline transfers, and ways of representing processing requirements and resources. (The last issue was raised in outline in the previous article I .) Other important aspects raised are data integrity and security, where the provision of clear unambiguous protocols to provide integrity and security for all users is a difficult problem. Program support is also considered - i.e. the definition of means for program creation, editing, compiling, linking, testing, storage, and associated file transfer and access mechanisms. It necessarily touches on programming languages and protocols that allow source and object programs and program modules to be transferred from one subwstem to another.
246
Other refinements to the model document included very careful consideration of the level of multiplexing (considered to be necessary at least two levels as indicated in the next section, and the position of encryption and data compression) and it was ultimately resolved that these should, for the time being, fit into the presentation layer.
PROGRESS AT THE M E E T I N G A considerable amount of common support was evident for a number of issues, such as the continued adoption of a top-down approach. However, a number of issues did not obtain unilateral support, for example, the proposal to process the model as a standard. This was not supported by
all those present, and was not fully resolved by the end of the meeting. There was also yew little discussion of a possible very basic model to allow rapid progress to be made, as such a document would have no obvious place within the ISO structure, and there would be no means for processing it. Other issues that generated considerable discussion were the question of definition of the role and position of the transport service interface. European feeling is that definition of such an interface is very important so that standardization above it and below it can be decoupled. In contrast, there was a feeling from elsewhere that definition of such an interface may constrain organizations active in the commercial sector, and act as a barrier to the provision of value added services by the PTT's and other line operators. Addressing within systems was considered by an 'ad-hoc' group, which came up with the recommendation that addressing is important within the higher protocol layers, and at least two levels of addressing are required - addressing of systems, and the entities within systems. Corresponding text will be written into the model document. By the end of the meeting, there were still a number of unresolved issues, which included the treatment of transaction processing, especially in terms of resources, priority, timescale allowed for turn around, and use of reentrant code (or, indeed, indication of requirements for such code). Other issues included the management of distributed applications, which has scarcely been considered, and the management of the standardization process itself. Liaison is still unresolved in terms of arranging positive two-way liaison with many interested parties, as also are issues such as off-line transfers. Sublayering has been passed to SC6 (its viability is questioned nevertheless it has been passed to SC6 for further examination). The interconnection of networks is still attracting attention, as also is the possibility of including a 'broadcast' feature that will allow messages to be broadcast throughout a particular
computer communications
network, or to selective terminals. Another issue not finally resolved at the meeting was production of two comprehensive lists of existing standards I which fit into the model, and existing or projected standards 2 which need modification to fit into the model. Discussion of these issues indicated that there is still a long way to go before standards can be drafted !
New work items It was generally recognized that progress is highly desirable, and a corresponding number of new work items were proposed, for issue by the secretariat of the Committee for postal ballot. Some ten items were covered the areas outlined below.
Reference model Further work is needed including progressing the present document to a point at which it can be considered as a potential standard. It will continue to examine, in particular, structural relations with in the architecture.
Physical layer The second work item corresponds to the lowest level of the seven layer model (the physical layer) and proposes that standardization of the services and protocols of that layer be undertaken. The major part of this task will be the identification of existing standards, and their testing against the OSI model.
Datalink layer The third proposal similarly recommends that datalink services and protocols be standardized, mainly by examining existing standards and testing them against the required functions.
Network layer A similar treatment is proposed for the network layer, although there
vol 2 no 5 october 1979
are fewer existing standards. The main standard which will need to be considered here (against requirements) is the X.25 standard for packet switching networks. Progress is the circuitswitched data network area is somewhat behind the packet switched area, and new standards may be required. After carefully considering these items in context, SC16 proposed that all the above items be assigned to TC97/SC6 (i.e. the data communications sub-committee) by TC97, if approved as work items. The remaining work items outlined below were proposed for assignment to SC16.
Transport layer The services and protocols of the transport layer of the reference model will need to be standardized very little existing work has been done in this area, and new standards are likely to be required.
Session layer In the case of the sixth work item, a layer that has only been defined since 1977 (as a consequence of SC16's work), new standards will clearly be required to satisfy requirements.
differing nature and size, including files in different codes and must include management and control operations.
Job transfer manipulation protocols Proposed as a subject for standardization, the ninth work item will be concerned with the transfer of jobs and related activities such as transfer of job output user information on job processing status, operator actions and manipulation of jobs, and the handling of enquiries on job status and the cancellation of jobs by operators and users. It will also include system message transfer. In this case,standardization of a virtual job control language (a difficult problem) is specifically excluded.
Management protocols The last (tenth) work item is concerned wi[h management protocols for dynamic control of mechanisms and parameters which are concerned with the management of open systems interconnection or distributed applications processing. They include protocols for system management (i.e. action such as startup, termination) and application management (e.g. data integrity, data authentication and data security).
Virtual terminal protocols An urgent need exists for virtual terminal protocol that will both fit into the OSl architecture, and satisfy the majority of user needs. It is proposed that virtual terminal classes be standardized, and the functions required to support each class be defined and appropriate standards produced.
File transfer, accessand management It is proposed that protocols he defined and standardized. These must conform to the OSl model, and will define an idealized virtual file store that will represent the properties of collections of file data, to be used for file transfer access and management operations. Standards here will need to support the use of files of widely
SC16's authority It should be noted that SC16 does not presently have authority to progress any work items which may be assigned to it, and it will need to have its scope (i.e. its working constitution) changed before it can start work on any such items. A revised scope was drafted for consideration at the November, 1979 meeting of its parent committee, TC97, which reads as follows: 'Standardization in the area of open systems as it relates to systems interconnection. This will include development of standards required for the reference model of OSI and exchange of information between open systems. This work will take into due account other standardization activities in this area'.
247
Providing the results of the (now due) postal ballot are known by November it is hoped that this matter will be seriously considered by TC97, and the UK has already requested its consideration by proposing a relevant agenda item for the TC97 meeting. In anticipation of certain work items being assigned to it, SC16 reorganized its working groups in line with the likely work it will have to undertake during the next two or three years. Working Group 1 concerned with the reference model still retains its old number, and will be concerned with its previous area of work, together with specific responsibilities for further development of the model, liaison with other SC16 working groups and other ISO bodies, vocabulary, formal description techniques, and principles of protocols and heading formats. A new working group, WG4, will be concerned with management of distributed applications, and have the title 'OSI Application and System Layers and Management'. It will be concerned particularly with the management of the application and system layers, data integrity and data security. The second new working group, WGS, will be concerned specifically
248
with the application and presentation layers and their contents; including file transfer and access management, virtual terminal protocol, and job transfer and manipulation. The last new working group (WG6) will be concerned with the session and transport layers, including all related services and protocols. Convenors appointed at the meeting were Hubert Zimmerman, reappointed to convene WG1, Dr A Langsford of AERE Harwell, UK, to convene WGS,and J McGovern, USA, to convene WG6. No convenor was appointed for WG4. In line with normal ISO practice all preceding and obsolete working groups were dissolved, i.e. the preceding Working Groups numbered 2 and 3 were dissolved as the new working groups were generated.
CONCLUSIONS AND OVERVIEW Although a final model has not yet been assembled, there was an optimistic mood at the meeting, and a feeling that work must now begin on standards projects that will fit into the final
model as and when it appears. As in the previous case, there was a good atmosphere throughout the meeting, and a cooperative spirit. Considerable progress was made, and it is hoped that this will be supported by the parent committee TC97 in November 1979, when it is hoped that it will revise the scope of SC16, and assign Certain work items to it, so that standardization can at last begin in this vitally important area.
F E Taylor* Systems Technology Consultants Knutsford, UK
REFERENCES Taylor, F 'Progress towards open-system standards' Comput. Commun. Vol 1 No 4 (August 1978) pp 210--217 Holdsworth, J 'Standards for open network operation' Comput. Commun. Vol 1 No 1 (February 1978)
* Leaderof the UK delegationto the London meetingof the ISO TC97/SC16Committee
computer communications