Do we really need another M ? - a question of standards

Reproduced here is an article by Keith Snell on M21, which is to appear in the German M-Borse and the UK MTA magazine.

The core of this article has been waiting in my word processor for an extremely long time. As the launch date for M21 approached I knew that we would be bombarded with the question "Do we really need another M?. It was a good feeling being ahead of myself, with the response already penned.

Since April 22nd we have had an enormous response from all over the world. Lots of old acquaintances have renewed contact, and no end of topics have been raised and opinions offered. Virtually every imaginable question has been asked.

Except the one that I had prepared for.

The satisfaction I felt on discovering that nobody out there doubted the need for another M was only diminished by the frustration of having to rewrite the article to an imminent deadline.

On reflection I see that the question is involved and leads onto the real issue that needs open discussion. Do we really need another M? can be discussed from both a commercial and a technical point of view.

Commercially, is there a need for another M? Do users want an alternative to that which currently exists, is there enough support for a company developing and marketing another M? Obviously the marketplace ultimately dictates. We have no doubt that with the technical and business background of the directors, M21 will succeed if it remains realistic in its aspirations.

Technically, is there a need for another M? M21 certainly does not claim to be a more all-embracing product than Intersystems' Caché. Nor does it claim to leave MSM, DSM and ISM back at the starting post. It claims to be a fast, stable, fully featured implementation of M that gives the M world an extra option.

Yet we believe that, if there is such a thing as the wider M community, then M21 is just what it needs. Such a bold statement arises not from arrogance but from the roots of M, the ANSI standard. We believe strongly that another M or two in the marketplace will effectively restore the standard and this will actually benefit M as a whole.

So the real discussion is whether a return to the standard is important to M.

For the youngsters in the M community, if that is not a contradiction in terms, a quick stroll down memory lane might help.

In the early days of M, when there were few implementations, the standards committee was formed. This allowed various companies to develop and sell their own implementations of M, confident that they all conformed to the same minimum standard. Some had more bells and whistles than others but essentially the products all shared this minimum functionality. Personally I look back on my time on the MUMPS Development Committee with an amount of satisfaction, feeling that despite criticism, the standardisation model worked well. For a non-commercial organisation, the MDC managed successfully to update the M standard to encompass hardware and operating practices not dreamed of when the MUMPS language was first developed at the Massachusetts General Hospital.

The problem may have been that M, not being driven by a single commercial organisation, just could not compete with the likes of Oracle and sections of the M world were successfully wooed by the big players.

Intersystems seemed to have the answer that would save M. Acquiring many of the other major implementers - Datatree, DEC and finally Micronetics, gave rise to an impressive organisation with a large customer base. With the promise of supporting individual implementations within the umbrella of a large, marketing oriented company it seemed a win-win situation.

So what is the problem and more importantly what is the answer ?

Over the years that I worked as Technical Director for Micronetics Europe I spoke to many people about the future of M and these discussions continued after the merger of Intersystems and Micronetics.

The fear was that when Intersystems finally achieved a monopoly over the M marketplace then Caché, their flagship, would be the focus of their development and marketing. Eventually users would be forced off their historic versions of M and into Caché.

Without a doubt, this is something to be welcomed by many M sites. Caché is a good product, being an M implementation it can compete with Oracle, Sybase and Informix. But not every M user wants to be part of this development. Very large numbers of existing M users do not need the full functionality of Caché. They would prefer not to pay prices that, though competitive with other database systems, reflect levels of functionality that they would never have chosen for themselves in an open market.

The fear is that, like Oracle users, they can be taken in any direction, and charged any price, with the only alternative being to abandon M altogether.

It can be difficult for organisations to commit to long-term developments if they feel locked into one vendor. It is true that this is the situation for Oracle users also, but this has not been the case historically for the M community due to the existence of a defined active standard and alternative vendors who supported that standard.

So despite Intersystems' excellent marketing skills, which have undoubtedly taken the company forward, M as a whole may not have benefited by the same degree. M has served a large community extremely well over the last 30 plus years. It would be wonderful for it to become more widely used, but long-term stability is far more important than competing with Oracle. Being the world's number one DBMS was never its brief and we would do well to question where we want M to go and what we, as users, really want of it.

The answer is very simple and I believe M21 is a key part of it. The answer is not necessarily to produce bigger, better and more functionally rich products, but to restore the standard and encourage new implementations of M. I believe that new implementations of M will emerge soon and with the open source versions such as GTM, there is already a basis for this belief.

Recently I spoke to someone who had been using CCSM for 11 years and claimed that at one stage there were many hundreds of sites worldwide using this version of M. There are other dialects of M out there still serving a useful purpose and now people will know that they can convert to another M when the need arises.

Is this 21st century Luddism? Not at all. Those who feel M21 does not have the features that they want or need can turn to Caché or another M implementation. Caché users who feel that they do not want to pay for features that they would never use can turn to other ANSI implementations of M. Users who have very low budgets can utilise open source versions.

The whole point of the standard is that a user can happily commit to any (standard) version of M and know that they can easily move from one supplier to another without prohibitive code or database changes. This certainly puts restrictions on the implementers. If they want to be regarded as M vendors, and speak for the M community, then they have to maintain that standard part of their product in common with other versions of M. "If it doesn't say M on the packet then it's not M in the box"

The choice then returns to the customer. All M users can commit to M knowing that they will not be locked into one supplier who can dictate the terms to them. This freedom of choice must make it more likely that M users will stay committed to M, and, of equal importance, the M alternative will seem far more attractive to Oracle users.

With the restoration of the standard and 21st century marketing, we could be in for a golden age for M.

Keith Snell
Technical Director M21 Ltd
(keiths@m21.uk.com)

M21 News