|
New Year 1998
|
Is there anything more mission-critical than running a hospital? The Medway NHS Trust, nursing an overburdened system with no upgrade path, undertook the Great Migration.
Why do companies and organisations choose to migrate from where they sit to a different system? The answer is progress. To become more efficient and hopefully save money. Every 3-5 years every MIS manager has to consider what’s going to happen next: do we re-write, upgrade the hardware or buy in a brand new system and all that goes with that. It’s a very big decision and one that can affect every part of a business or in the following case the running of a busy hospital.
In 1993 Medway Hospital, now The Medway NHS Trust, had just that decision to make. At the time they were running a generic Pick system from Fujitsu, which was ‘busting at the seams’ with zero upgrade path. Apart from experiencing severe file corruption on a weekly basis the system was slow and work procedure was planned around the computer’s behaviour. Any report of any size had to be run at quiet times, overnight or over the weekend. Even system backup had to be scheduled when fewer people were about. The system was inadequate to meet the needs of the business, in this case the running of a major part of the hospital.
The software was mostly written in-house and comprised approximately 20 applications
covering every aspect of hospital life from Accident & Emergency to an internal
Help Desk to support the end users around the hospital. It was written in Pick
BASIC, PROC and ACCESS. Extensive use of Multi-Value fields and DICTionaries
was evident.
There were four main criteria for the new system:
The evaluation team spent several weeks evaluating Pick emulations on UNIX and one other product new to them called jBASE. jBASE stood out from the crowd as it was not an emulation engine on top of UNIX yet it enabled 100% of the applications to run as real UNIX applications. If the platform of choice was Windows NT then the same would be true of NT in that jBASE compiles BASIC to .exe and .dll’s. The reason jBASE stood out was the unique and skilful thought that went into its design. Rather than run as an environment on top of an operating system the developers implemented a method whereby the BASIC applications could be transparently converted to C source and then compiled to machine code by the UNIX C compiler, or on NT the MS Visual C++ compiler.
The immediate effect of such an enterprising trick was improved performance several times that of its nearest competitor. During pre-purchase tests it was declared that a report had failed to run under jBASE, but on investigation it was found that the report had completed 30 times faster than on the existing system. The results checked out and it began to sink in that this jBASE stuff was something else. With such speed improvements, also assisted with access to the then latest processor power meant that the MIS department could manage its daily tasks in a more natural, more efficient manner. They could inform their end users that on-line real-time reporting was available and they could thankfully back up the data to magnetic tape in a civilised manner. Further, new and even more complex reports could now be introduced without fear of the system grinding to a halt.
Although speed is of major importance it is in fact almost a by-product of the design and techniques used in jBASE. A far greater benefit is that the legacy applications become real UNIX (or NT) applications and are therefore able to interpolate with other UNIX (or NT) applications. jBASE has its own database independent driver, a piece of ‘middleware’ called jEDI, that divorces the BASIC application from the database. In other words the BASIC program, without change, can write the data to non-jBASE databases simply by introducing a pointer to a jEDI to a DOS file or Oracle for example.
jBASE obviously comes with a MultiValue database jEDI but also a DOS, UNIX and NT file driver too. Other drivers for any database can be added to any database depending on the application needs. Multiple databases can be simultaneously searched, processed and updated from the same front-end BASIC program! All this is wonderful technology but for The Medway NHS Trust on DayOne their main objective was to run the software exactly the same (but faster and without falling over) as they did before. They needed zero end user re-training on DayOne. This was successfully achieved and with the transparent openness of jBASE other activity could take place at a later date as and when the MIS department wanted to.
The latest arrival to the jBASE technology family is jBASE OBjEX. This makes Visual BASIC look like the BASIC MultiValue developers are used to. Therefore the learning curve for existing BASIC developers is minimal - not to mention the sheer power that blending Visual Basic with jBASE provides. Another feature of jBASE OBjEX that will affect the business performance is the ability to access the MultiValue data directly from Word and Excel and other VBA applications. This kind of integration has only traditionally been available to the mainstream; as jBASE compiles the legacy BASIC to be a mainstream application such natural integration becomes a reality.
The proof of any technology is in its use. jBASE went live at The Medway NHS Trust on January 3 1994; when asked about downtime as a result of jBASE no-one at the Hospital or JAC could record any at all. Suffice to say jBASE has proven itself in the most mission-critical of sites possible, a hospital! jBASE maybe the new kid on the block but JAC have announced the 10,000th user, a figure that looks set to double by the end of 1998. A technology of today and the future.
John Taylor – Freelance Reporter
Last Updated: 29 July 1999