The Old Joel on Software Forum: Part 3 (of 5) - Foundation for large business software

Foundation for large business software

For little business purposes it seems to be clear. Microsoft gives there very good tools and technologies. Access fits most uses and is expendable through COM-components. It's proven as there are out a lot of those solutions and one can do it very rapid. Let's say such system is for one user capable and if you also implement transactions I guess up to ten concurrent users have a real advantage by such a system.

As systems get bigger, as they always do, things begin to become less clear. Of course the Access solution can be expanded by a database server like SQLServer for lager data and faster access. But integration, communication and costs start of to matter.

Integration in itself is complex. But let's simplify and only deal with the integration of various data sources and various subsystems such as a compuer connected scale.
The communication issues arises most from integration. Since the software doesn't work anymore alone, it has to communicate with various systems like the former mentioned scale or a transactionmanager for distributed data access and manipulation. And of course, as it deals with business demands, costs are probably number one demand. Licencing and the increased developing effort rise costs. Rapid development should compensate this.

So my issue is, on which foundation should one build e.g. a warehouse application for simpler discussion for let's say up to 500 concurrent users and various databases and subsystems rapid ?

Michael Bruckmeier
Monday, October 13, 2003

If I understand your question, I would recommend Unix/Linux based platform, corresponded available software and Java/C++ for development. (But not sure if that your question...)

Evgeny Gesin /Javadesk/
Monday, October 13, 2003

One more comment. For the EAI (enterprise application integration) I recommend the IBM MQ Series, Workflow and Integrator, if these tools suit your business needs. Of course your budget should allow them.

Evgeny Gesin /Javadesk/
Monday, October 13, 2003

The answers go that way, but I would like to hear of some concepts like I had such a project and we used this and that. I want to figure out some sort of common way dealing with this requirements.

Michael Bruckmeier
Monday, October 13, 2003

Stand on the sholders of others:
http://www.compiere.org/

Doug Withau
Monday, October 13, 2003

"Access fits most uses and is expendable"

I agree it is expendable. Whether or not it fits most uses is another question.

Milton
Monday, October 13, 2003

Evgeny, you are so right.

I've now survived fourteen projects in the EAI space. The ONLY projects which were an unqualified success were those based on MQSeries.

Vitria sucks. Tibco sucks the worst. BEA is trying to suck, but not quite there yet.

Frankly, I think the old adage of "you won't get fired for buying IBM" might still be true. If you need unattended mid-range, buy the AS/400. Need larger? Buy AIX/Linux. Need the largest? Well, mainframes will STILL be here for many years to come. No matter what, use a foundation of DB/2 and MQSeries, and you'll never have to make an architectural u-turn.

HeWhoMustBeConfused
Monday, October 13, 2003