Database Technology

As stated in the text of “The development of ITC infrastructure at UU and SLU after 1960” (1965-1970), the concept of a university computer center included installation and operation of a powerful electronic computing machine facility. In addition, the concept included as well the availability of a so-called Program- Development Support Group, PDSG, with the task of initiating the use of such a facility in other academic areas than – as traditionally – natural sciences. University computer centers had, in addition, the task of using this PDSG to spread awareness about the new technology to both the regional industrial, and regional public sectors and – on commercial terms – helping them with the use of this new technology.

It was this part of the infrastructure that laid the foundations to many of the essential and enduring innovations that have their roots in Uppsala. Mimer is one of these.

In the late 1960s, there were several machine capacity-intensive application programs in routine use in Uppsala. The PDSG had developed these together with different users. Some examples are systems for public health control and administrative systems for UU, SLU and other universities in Sweden.

At that time a major obstacle in all initial and further development of systems was that each application program needed its own way of storing its input data. Other administrative support systems could, in other words, not use the data comprised by e.g. the computer based staff wage payment system.

At UDAC Werner Schneider and his closest collaborators became increasingly aware that the storage and retrieval of data should be effectuated through a special system that could be used by any application program or system. However such a system was not available.

When, in 1971, UDAC took a new computer system, the IBM 370/155, into operation, the PDSG, along with other users, attained access to such a system, the IBM IMS. This represented a clear step forward, albeit with many traps, obstacles and disadvantages. For UDAC, the primary disadvantage was inefficiency. IMS was a true innovation, but terribly inefficient with regard to computer time consumption.

One of the program/systems developers at UDAC, Åke Persson, found in the book “Systems Analysis for Data Transmission” by James Martin, a chapter that dealt with relational databases. Consequently, he also read the original publications by E.F. Codd, a researcher at IBM, the originator of the model for this kind of databases. (See “A Relational Model of Data for Large Shared Data Banks”).

On the initiative of Åke Persson, Werner Schneider formed already at the beginning of 1974 a development section within UDAC primarily to develop a prototype for a relational database, RAPID. To achieve the best possible efficiency and memory utilization, RAPID was written in IBM assembler.

The first version was put into operation in various application systems from 1974 to 1975. As a Canadian computer company wished to register the brand name Rapid for a similar database system, UDAC’s RAPID was renamed in 1977 to Mimer, in order to avoid an unnecessary trademark battle. It must be said that this was a clever choice of name. Mimer was in Norse mythology a giant who guarded the well of knowledge and by drinking from the well, he was able to answer all questions. None other than Mimer had access to this source, but Odin sacrificed one of his eyes in the well and thus could look both forwards and backwards in time. One could say that Mimer was the Gods’ information processing system. In addition, Mimer’s well is, according to legend, in the Uppsala region.

(PICTURE: Mimer drinking out of the well)

It wasn’t long before other parts of universities, including other university computer centers, also wanted access to Mimer. As these centers used computer systems, such as Digital (DEC-10, PDP-11, VAX), Norsk Data, Prime and Control Data, Mimer needed to be adapted to these systems. To avoid reprogramming for each target environment, it was decided to make Mimer machine independent (portable) as far as possible. It was therfore programmed using the most rudimentary core of the programming language FORTRAN. Less than 2% of Mimer was machine dependent, and thus had to be reprogrammed for each computer system. The remaining 98% was portable between all the different types of computers. The idea for this had its roots in the experience gained in the period

1960-64, when researchers – among them Werner Schneider – had to pick up machine time on any computer anywhere in the U.S. and Europe to run time- intensive programs in various research areas. The task was, in other words, to define the common core of FORTRAN for all types of computers.

Mimer had two modules: one was the database management system, MAMI (Mimer Access Method Interface), and the other was the query language Miman. Around 1980 Mimer was supplemented with the 4GL-tool Midam. This was a program generator that from a high-level representation could generate MAMI applications in Fortran and Cobol. Midam was based on Tore Risch’s doctoral thesis Lidam.

With the help of Mimer, database applications could be made transferable between different computer environments, but the presentation parts had still to be reprogrammed for every specific computer environment. Therefore, at another department of UDAC, a portable display manager, UDAC Screen, was developed, which also became part of the Mimer family.

In order to better promote the various modules of the Mimer family, a name change occurred in 1982:

MAMI – Mimer/DB
Miman – Mimer/QL
Midam – Mimer/PG
UDAC Screen – Mimer/SH

The Query Languages module was called “QL” because it was chosen to implement one of the most widely used international query languages, QUEL.

A major breakthrough occurred on Very Large Databases, a World Congress of database researchers in Rio de Janeiro in 1979 where relational databases were the main topic. UDAC was represented by Åke Persson. One of the presentations described a completely new transaction management approach, “On Optimistic Methods for Concurrency Control” by HT Kung and John T. Robinson (http://www.seas.upenn.edu/ ~ zives/cis650/papers/opt-cc.pdf). This method is based on the fact that conflicts occurring due to simultaneous database access are

unusual even in congested transaction systems. Instead of pessimistically locking database resources at each database access, the Optimistic Concurrency Control- based system deals with conflicts when they arise. This makes the system both more efficient and convenient.

At the suggestion of Åke Persson a new version of Mimer was developed in which this new transaction management method was implemented. It was a huge success! None of the competitors on the database product came close to Mimer’s performance.

Even Mimer’s query language module was changed significantly. The IBM developed SQL language was the only query language that has been the subject of an international standardization process. This meant that Mimer had to replace QUEL by SQL.

Both of these changes laid the foundation of the product MIMER SQL still 2012, is the world’s most advanced relational database management system. It is the only system that is based on optimistic concurrency control. MIMER SQL is available for Microsoft WindowsMac OS XLinuxAndroidSymbian OSUnixVxWorks and OpenVMS. [1]

Mimer belonged to UDAC until 1984 when it was transferred to a newly established separate company Mimer Information Systems, with Sven G. Johansson from UDAC as manager. The company still has its headquarters in Uppsala, which is now named Mimer Information Technology. All development and international marketing and sales are handled from there.

An example of Mimer products can be found in the excerpt ”Mimertillämpningar” or “Mimer applications”.

By Stefan EckMimer SQL Marketing.