DSXchange

We are currently looking for an ETL tool that meets our requirements, and DataStage is one of the front runners. Why I ask about Pick is that I have 20 years of experience in PICK, and was wondering if the latest version of DataStage is still running under UniVerse. While we are primarily an Oracle shop, we do have a group that supports our billing system which runs under UniData (I started with the company in this group, but switched to an Oracle group over a yeare ago), and I was hoping that having a team of PICK programmers would be a plus for choosing DataStage. If DataStage is still running under UniVerse, I would also like to know how helpful knowing PICK is in implementing it. One of our requirements is the ability to load massive amounts of data, so we will most likely be running PX. However I've read that PX uses a different architecture.

Any opinions are appreciated.

Thanks,

Rich Sorice

No.

Your BASIC and/or Universe knowledge will help you when working with Server jobs. Technically, DataStage Server jobs now run under the "DataStage Engine" and not Universe as there has been a code-split and they are actually seperate products with a common ancestor, but for all intents and purposes they are still the "same thing".

As for PX, you are correct in stating that it is an entirely different architecture, based on the Orchestrate product from Torrent which Ascential acquired or licensed. While you can use BASIC in PX via a special Stage, the language of choice there is C.

They are two completely different animals.

Thanks for the info. I knew all about DataStage's anscenstry (UniVerse=>Ardent=>Informix=>IBM/Ascential), but wasn't clear about PX. My concern is that with the large volume of data we deal with (terabytes), we would want to start off using PX, negating any benefit of knowing PICK. I guess we could still use DS for smaller jobs.

Thanks,

Rich

You'd be surprised how much data you can shift with server jobs, particularly if you use the multi-instance capability.
While this can't partition processing across an MPP cluster, it can take full advantage of multiple CPUs in an SMP environment. The only downside, compared to PX, is that you (the designer) - rather than the engine - must specify how the data are to be partitioned, most likely with job parameters.

What Ray says - don't write off Server jobs! There are some tricks and the burden falls back more on the designer of the jobs than letting it be handled automagically by the engine, but you can move some goodly amounts of data with it.

Thanks, thats good to know. I was under the impression that DS was single threaded, and would only run on one CPU in any multiprocessor environment.

It makes a lot of sense, then, if all else being equal, that we use some of our existing inhouse knowledge rather than training people on Informatica's programming language.


Rich

Besides, if you went with Informatica, you wouldn't have... well, us!

True.

I just have to convince the other 6 groups involved in the search that have never heard of U2 or Pick!

Ray, your name sounds very familiar. You were involved in Oliver's U2 list a while back, weren't you?

Ahh, the good old days!

Yes, I have three hats.
And like to stay active on the various forums for each, mainly:

DataStage (here (where Oliver's DataStage is archived) and ADN)
UniVerse (www.u2ug.org, which took over Oliver's U2)
Red Brick (www.iiug.org)

PX and server are yes different animals, but you job control with PX can still be written in BASIC. It is prfeerred that you use the job sequencer which will write it for you. The DS Engine is still required for for keeping log files. job information etc as in server jobs, so our knowledge would still be an advantage.

Only 20 years, I remeber being part of teams that wrote virtaal run machines, the fundametal difference between PICK is that PICK was the OS. Wherea PI, UV and UD and DSEngine are hosted by the OS. So whilst the functionality looks he same, the implementaion, totally diferent.