I've been asked to give you my thoughts on the current state of Pick. The Pick system has been around officially for just over twenty years. It's hard to believe its been as long as that, but its true -- the original Microdata Reality systems were first shipped in 1973. In the nineties, when there are dozens of portable systems many people fail to realize that in fact, it was Pick who introduced the concept of portable operating systems. Pick made portability a commercial reality (pardon the pun) when UNIX was still an academic curiosity.
Pick's original strengths, those that allowed it to prosper in the early years, are still some of its greatest strengths. The primary original strengths were clearly the significant performance advantage Pick enjoyed and its ease in programming the simple business solutions required for the time. In the seventies there were no PC's and very few mini-computers. The world was governed by main-frames which cost millions of dollars each. Most of these computers ran what we would consider batch operating systems. Pick was welcomed into this void. An early Pick computer could put 16 to 32 terminals on-line for about four to five thousand dollars a seat. It ran three to five times faster than the few proprietary mini's available. It was easier to program than its predecessors and was also incredibly forgiving of late design changes.
This speed and programming ease in turn, created the asset that drove the Pick system to success in the late seventies and all of the eighties. The asset I'm speaking of is of course the large pool of Pick applications available worldwide. As Pick was ported to an increasingly large pool of hardware, these readily available and reasonably mature applications were clearly the life-blood of the Pick growth phenomenon.
Of course, life is never perfect. Pick was never designed as a formal data base. It does not satisfy many of the 'relational rules' and was not designed to be 'bullet-proof'. The very wholeness and single source mentality that made Pick efficient in the old days is a detriment today. Today, Open Systems and Open Solutions are the required coin of commerce. You must allow users to site data and applications on the hardware and databases that make the most sense and then integrate the results into a cohesive solution. The traditional Pick vendors have been slow to develop improvements in these areas.
New comers (at least relatively) have been much more aggressive at peacefully co-existing in Open Environments. At the low end, PC based databases such as Microsoft Access, Dbase, Foxpro and the Pick-like Revelation have all made significant inroads in the low terminal count, first time data processing shops that have always been a major stronghold for Pick. While these types of systems usually lack strong transaction integrity protection, they are fairly competitive in other areas especially in the presentation or GUI area. Also, the hardware and networks supporting them have become reliable enough that reliability is generally not a serious issue for small shops. All they seem to lack are the pool of basic applications Pick has. I expect them to easily develop them as time goes on.
At the other end of the spectrum we have the formal relational data base vendors such as Oracle, Sybase, and Informix. These companies are all very large companies and need to be taken seriously. They have solved all the reliability issues and are working on the distributed data problems now. Their databases and tool sets seem to allow them to create large, complex systems and, if newer design methodologies are followed, to maintain them. Fortunately for the MultiValue community, all their formalism does not (yet) add up to truly easy application development -- case shows tremendous future promise, but the hype still seems to ahead of the substance. Also, their raw performance is low enough to leave the Pick applications an edge to work with. While I feel it would be foolish to dismiss competitors of this size and staying power, for most of us, they are still future threats (or options, depending on your viewpoint).
That brings us to two really interesting alternatives -- Vmark and Unidata. Both support newer implementations of highly Pick compatible data base engines. Both now have full formal transaction boundary protection. And both have new GUI options. But there are differences too.
Vmark's uniVerse is a 'C' native implementation of a database engine originally modeled after Pick's. The full 'C' design allows access to the whole range of UNIX utilities and add-ons. Also, their acquisition of the Computervision PI/open product in late 92 gives them a much larger market to work with and some very advanced technology. Calling their database a "Post Relational" structure, they are attempting to provide a highly compatible migration path for traditional Pick customers while still appealing to non-Pick database developers. A rich set of offerings includes many new products in the "middleware" arena. This term is used to describe tools that fit between the database engine and the user applications and provide expanding functionality such as GUIs or client server features. At three time the size of Pick Systems and accelerating, they are long term players.
Unidata has been the dark horse of the race. While starting a couple of years after Vmark, they have been aggressive and competent competitors and now appear to be close to Vmark in total volumes. I think you'll see a lot more of them in the future. They are the only company in the Pick market with an entirely new and different database implementation. They had a more formal start with a SQL compliant engine modeled not on Pick per-se, but a mathematical model they call the "Non-First Normal Form". This engine is structurally suited to distributed processing and is able to support Pick values and sub-values. It is also built as a 'C' native product allowing access to UNIX tools. Their strategic alliances with the outside database vendor Sybase, strong developmental relationship with System Builder and recent acquisition of Wintigrate as a first class PC based GUI are all interesting examples of not just good marketing, but true Open Systems thinking as well. Together, these two alternatives are winning the majority of the replacement business in the Pick marketplace. Both are also attempting to penetrate other markets with their ease of use and Open Systems outlooks.
How does this whole situation stack up for the current end-users in the Pick market? Well, I think it paints a very positive and encouraging picture. To recap the material above, the Pick database, in its many forms and vendors, has not only survived the last twenty odd years, but it has prospered and is still somewhat ahead of the outside "relational database" vendors in most areas. If there was any question about the correctness of your historical decision to go with a Pick-like database, you should easily be able to hold your head up and defend your decisions.
As to where the market is going today and where you should be going, I think the outlook is also positive. While many of the smaller and weaker vendors have fallen by the wayside, they have been replaced by larger and clearly better managed companies. The trend to looser ties with the hardware vendors has directly increased the ability to pick the database you feel is best suited to your application and future and to separately find cost effective hardware to run it on. This is a classic Open Systems benefit the Pick market has always had and that has really accelerated over the last three or four years.
If you are a small endures and your application is already satisfying you needs, there is no reason to take any significant steps. You probably have little need for the added options offered by the newer vendors, and there will be traditional vendors with high quality hardware at decreasing prices for the foreseeable future. In fact, if your hardware is more than three or four years old, you might take a quick look around -- prices and maintenance costs have dropped so fast recently that you might find you can improve your return on investment with a simple swap to newer hardware.
If you are a bigger endures and have needs for SQL, distributed processing, PC based GUIs, or any of the other features the new vendors are showing, take a good look at them. This goes for the VARs and software houses too. These database vendors (especially Unidata and Vmark) are well run, financially solid companies, building platforms that will enable you to penetrate increasingly complex marketplaces with your current applications. The conversions may still have some rough edges for some people, but this will continue to get better over time and there is absolutely no need to scrap your applications and start over. Yes, you will have to add some energy to develop GUIs or to take advantage of other new features, but you can do it in a safe, incremental and structured manor.
In conclusion, whether its called Pick-like, Post Relational, or Non-First Normal Form is perhaps not as important as the fact that your software investment is safe. It will have a world class database engine under it to enable it to continue to grow and expand into the next century.