2960 DME S4
Title: OUTLINE DESCRIPTION  Section: 2

2.0

DME/S4 is an implementation in microcode of the total System S4 order code together with modifications to J and Multijob software to provide an environment on 2960 in which System S4 object programs can be run. In that the System S4 order code and software are interacting directly with OCP, store and peripherals, no 2900 order code is entered and concurrent running of 2900 jobs within this "System S4" environment is not possible.

Providing that sufficient equipment is available within the configuration, 2900 work to be run under VME/B or VME/K can be run on the hardware when the correct 2900 microcode is loaded. This will of course be non-concurrent with DME/S4 operation. Similarly 1900 microcode can be loaded for non-concurrent running of DME/1900 The distinction between DME/S4 environments and the System 4 microcoded emulation under VME/B (S4BMEEP) environment is best illustrated by diagrams of the software structure. (Fig. 1 & 2)

DME/S4 + S4BMEEP

With DME/S4 only System 4 software and System 4 microcode are supported but in the case of S4BMEEP both S4 and 2900 software and microcode are present and the S4 work is carried out within one VM under VME/B coexisting with other 2900 VMs.

DME/S4 performs at a faster rate than S4BMEEP because, although similar System 4 software is used in both cases, the path lengths for 2900 I/O operations through the specialised microcode are shorter than their logical equivalents through the Emulation Service Routine and VME/B.

Overall performance of DME/S4 depends on processing capability and peripheral driving capability. The processing capability of DME/S4 is targeted at 80% of a System 4/70 and this together with faster 2900 peripherals makes it a useful upgrade from a System 4/50 or 4/52 processor system and an effective, short-term replacement for most System 4/70 and 4/72 processor systems on multiple sites, as part of a defined transition path. In the case of single, heavily loaded system 4/70's, additional capability would be required over and above that provided by a DME/S4 configuration.

An extensive range of 2900 and System 4 peripherals is supported. DME/S4 supports a major subset of the System 4 peripheral range. This is not true in the case of S4BMEEP because VME software has not yet been written to support System 4 peripherals via the GPC.

DME/S4 supports the two major System 4 software environments, namely J + CCP/Driver and Multijob. The System 4 range of superstructure products (e.g. compilers and utilities) will therefore run on DME/S4 as on a true System 4 machine but with a small number of facility restrictions as detailed in section 14.1

DME/S4 is another step in ICL planning to ease the transition path from System 4 to 2900. Users can now run their System 4 production work without a need to know the 2900 operating systems and other software. Their work can continue within the known confines of their current J or MJ environment. On another shift they can begin development work with 2900 microcode and VME/B or VME/K as a separate exercise, extending it to the use of S4BMEEP (Where VME/B was chosen) if necessary.

Thus the training and support costs implied by the introduction of new hardware and software can be spread over a longer timescale compared to other methods of transition.

To summarise, the options for transition now available for System 4 users to 2900 are vastly expanded.

MARKETING THE 2960 DME/S4

3.1 Introduction

The 2960 DME/S4 product has been developed to provide an alternative means of bridging from S4 to 2900. It allows a "double bridge" approach with a consequent spreading of transition costs and resources and a reduction of the risks associated with a straight transfer. As a by-product it can also be added to an existing S4 site where the S4 equipment is inadequate for an increasing workload.

Its existence will help to preserve in the short/medium term the System 4 Customer Base which would otherwise be at risk because of the cost and time involved in direct transition to 2900, and the limitations of the emulation aids available. It will also result in increased 2900 hardware revenue in the shorter term.

The product will be sold as a standard 2960 system albeit with the DME/S4 option.

3.2 Strategy

Basically the product must be used as a means of obtaining 2900 orders and revenue early. It must be marketed as a temporary device which provides a facility allowing the customer to get across to 2900 quickly, cheaply and safely. The customer, once across must be encouraged to move towards true 2900 mode, and develop new applications, with J 2900 software, for program and system trailing. In essence the customer can do all his system replanning, redesigning and rewriting after the new hardware has been installed, instead of before. This means earlier orders.

Although DME/S4 provides communications capability, it must be noted that in view of the comprehensive facilities provided under VME, comms applications will be the first that the user is encouraged to transfer.

3.3 Customer Profile

The average S4 customer is technically competent, has a large system, much of which is written in a low level language and often incorporates communications. Many of these systems are also tape orientated. There will be a wish to move on to the 2900 range to take advantage of its extra benefits and also to avoid the problems of potential deterioration of ageing equipment, with a consequent lowering of performance.

The customer will, on the other hand, wish to protect the heavy investment represented by his programs, particularly those which are of recent development and implementation and are stable, or are run infrequently. He may be disturbed at the high costs of direct transition to 2900 and may be considering emulation as an interim.

In summary he will be looking for

Security

Continuity

Optimisation of Investment

Performance

Economy

Individual customers will have other major requirements in mind but these five are common to all.

Section: 3

User Benefits

3.4.1 Security

The existing systems continue to run as before and can be maintained or updated as before where this is necessary. There need essentially be no change to operating procedures, files, coding or end-user interface except that the throughput may be improved by the faster peripherals and transfer rates. New software development can take place in a separate time slot in a 2900 VME environment without impairing the production work, so allowing planned evolution to True 2900 working. This plan is under the control of the user and governed by his timescales, his resources and his commercial or organisational requirements, and can be modified to meet changing circumstances. It is not governed by external machine or system considerations beyond his control. If they are not available on site, standby facilities may be organised with other S4 users.

3.4.2 Continuity

A smooth and effortless move from S4 to 2960 DME/S4 will ensure continuity and avoid the risks inherent in moving new systems to new equipment with new operating systems all at the same time. The user is free to convert his applications to a 2900 software environment at his own pace within his existing resources and with the use of his own machine testing facilities.

The move to 2960 DME/S4 will also enable the user to plan the training of his staff in 2900 and system facilities over a period, rather than having to acquire a new range of skills in a short space of time before taking delivery of new equipment.

3.4.5 Optimisation of Investment

Firstly, existing systems represent a very considerable investment in terms of resources and time. Many systems will be adequate for the needs of the users over the next two or three years, and some programs will be run so infrequently that it would be uneconomic to rewrite them in the shorter term.

The 2960 DME/S4 allows the user to continue with his present system until his business needs dictate a change or replacement. Secondly, many users will have peripheral equipment which is well within its maintenance life. As the peripherals listed in Section 7 can be connected, 2960 DME/S4 gives the user the opportunity to optimise his investment in this equipment. All this means that the user' s options for future development stay open while the work is transferred and he is not restricted by a general system freeze as would be the case if converting to, or rewriting for, a new range of equipment.

3.4.4 Performance

The hardware performance of the 2960 DME/S4 is extremely competitive in that its basic OCP power is treble that of a System 4/7O. In particular the processor system has a significantly lower hardware fault rate than that of the S4-70/72.

At the same time both J and MJ are stable, efficient and fully developed operating systems. The combination of 2960 hardware and S4 software will therefore make for a very reliable product.

Finally it must be remembered that much of the S4 peripheral equipment will be nearing the end of its contractual maintenance life, after which it will cost more to maintain and have a lower contractual level of serviceability. Because of this, and the problems of maintaining obsolete equipment, the customer must always be encouraged to replace S4 equipment by 2900.

3.14.5 Economy

 3.14.5.1 Resources

When transferring normal programs from S4 to 2960 DME/S4, no application programming effort is needed. System regeneration will of course be necessary but no more than after a hardware enhancement or than when moving to another S4. User resources will therefore be available for the development of new systems and conversion of existing ones. These resources will also be available for reorganisation of the files onto disc within the existing S4 environment if this is required.

3.14.5.2 Cost Effectiveness

The 2960 DME/S4 shows a very marked cost/benefit advantage over S4 machines. Its processor power is over three times that of a S4-52.