Strategies of use for webMethods Integration Server

Hello erverybody,
I have to establish an criteria catalog to reach a decision.
It should help to decide in which cases it makes sense to conect to applications through webMethods an when is a point-to-point connection better.
I appreciate for any suggestion, experience, tips and tricks, etc.
best regards
chris

Hi Chris,

I had a similar request from one of my customers - when to use different integration patterns. The answer really depends on the requirements and objectives that you are trying to achieve.

Since you have access to Advantage - I would suggest that you review some of the GEAR documentation that could help you with this topic.

Start at looking - “GEAR 6 Architecture Approach and Guidelines.pdf” - it details the different architectures that wM Fabric can be use for.

http://advantage.webmethods.com/bookshelf/Best_Practices/GEAR_6/Overview/GEAR_6.zip

In my opinion there are multiple reasons to use a message broker/integration server when integrating two (or more) applications, even in a p2p integration:

  1. interfacing is always done using one single technology (skillset) that is fit for the job
  2. gives the possibility for future reuse of application adapters
  3. when using hub-spoke models based on a enterprise canonical then using the IS is almost implicit ofcourse

Ofcourse, introducing an extra platform/technology to the equation will affect operational maintenance and costs aspects but I am sure these will be compensated by reasons 1 and 2.

In the end, it all comes down to strategy I guess…

Hello,

thanks for your help.
My problem is quite a bit more specific. You have to know I’m a student apprentice, so I have less experience.
The computing environment of the company is heterogeneous. We have the Integration Server in use with some applications. And now it is my task to create an criteria cataloge to decide: when makes it sense to integrate legacy systems / new systems and when should we buy an adapter to connect applications direct without wM.
best regards
chris

I personally believe that all systems should be integrated through a central bus, provided that access to these systems is required in the same manner. Since WM integrates with just about everything (some easire than others via supplied adapters), then I think the integration of these systems is more a timing/rollout issue than a “which should I pick” issue.

Some criteria I would consider would be:
Ease of implementation
Speed of implementation
Performance hits or improvements
WM provided adapter/integration
Custom integration

…something to think about…

…if in-doubt just Integrate.

Create re-usable components (for legacy and new systems) that sooner or later you’ll find a consumer for it.
It’s part of the “soap-box” material for a SOA environment.

Have fun…

Nino.

I suggest, its based on No. of system systems and transaction types involved between those systems… and data load.

Based on this you can decide really need to go for WM hence publish subscribe methodology.