Need to restructure mercator

[mlapse]

Hi,

My company is in the process of reevaluating the Mercator architecture we have. It is in a very shoddy state. Ex: there are no systems at all..only run maps which are called from various places and make the visibility very poor. Also Commerce Manager is not being used at all.

i am looking for any information on how to correctly design and implement Mercator in any environment. Basically we are looking at a scalable and extendible solution which reduces the current prod support time and provides a good overall solution.

Any suggestions would be helpful.

Thanks in advance

 

[JuJutsu]

What are your trying to do with the product? What version are you running, on what platform?

The DataStage TX product is very flexible, you can probably find 4 or 5 different ways to get the same output.
If you can upgrade to the new 8.0, it has a map profiler (to find out where a map's performance might be tweaked, a map debugger to find out why the output you expected is not being created, and a page size and count tuner for better performance. As to overall design of a system and specific maps, this is more of an art than a science, and an experienced, competant DataStage TX architect would be the best source for answers. You might want to contact IBM support and ask for sales to contact you with any white papers they might have.


JuJutsu - Jeff S.
Support Analyst

 

[eyetry]

While we use TX and Cmgr here, we've had similar issues with cost verses return. Over the last 12 months though, I've noted a more proactive use of the tools. I think the key is involving the right people and having a clear vision on where the tool fits into the big picture.

Here, we receive similar data from numerous external sources most have unique formats. Some of the inbound/outbound data formats are fairly complex. So, we map all external formats to one internal per data type. It saves a large amount of custom coding, testing etc, data format issues, reduces abends and so on.....(When we first implemented the processes two mainframe developers said that our maps reduced their development and testing by 30-40 percent and our data center said that the scheduling notifications that we started sending made parts of their job easier too.)

Problem here was visibility, no sizzle, savings were soft dollars and hard to quatify. At the same time we brought in another tool specifically for printing and mailing and there were lots of hard $ saved.

I agree on the upgrade to 8.0. Saw the beta info and some of the features are pretty nice, like the debugger and also the profiler. Very nice!!! We'll probably wait for the first major patch before implementing though.

If you participate in Developer Net (still developernet.ascential.com) there are some white papers that discuss deploying a TX environment.