Do Informatica skills translate to Datastage skills?

rhale · Post by **rhale** » Mon Mar 03, 2008 12:41 pm

I am interviewing people for a couple of positions in our company and our recuiters are having a hard time finding people with Datastage skills. So we are concidering interviewing people with ETL and Informatica skills that have express an interest in learning Datastage. And while I feel comfortable in evaluating a persons Datastage, SQL and UNIX skills, I am not comfortable in evaluating their Informatica skills. I am also concerned about how easily Informatica skills transfer to Datastage.

Does anyone want to share thier experiences going to Datastage from Informatica? How difficult is that transition? And are Informatica skills useful in a Datastage environment?

Thanks for your time.

kcbland · Post by **kcbland** » Mon Mar 03, 2008 12:55 pm

You're correct, it's a difficult comparison. Since Informatica keeps programmers from the low-level coding and reinforces doing things the way the tool works best, Informatica has gained a reputation for being easy to use and develop.

Fanatic programmers love DS (Server and PX) because of the awesome capabilities to deep dive and write from the lowest level up. The main criticisms of DS seem to be its difficulty in use and its infinite capabilities.

I've worked with programmers who were very business savvy and they disliked DataStage because it seemed too technical. Vice versa, I know programmers who hate Informatica because the "best practice" way of doing things didn't work well with Informatica because of its framework.

I think you can teach anyone either of the tools easily, the question is do they have a head on their shoulders? I often ask candidates do they know how to monitor server cpu, disk, and memory utilization to more effectively measure the throughput and performance of their job. With Informatica background developers they never need this skill. So when they're writing shell, perl, or c programs they tend to be lite on these skills. But DS folks are more apt to be comfortable working within the operating system environment. They tend to be more low level coders and refer to themselves as such.

I'd say the difference, in my opinion, between Informatica and DS folks is the difference between mappers and coders.

ray.wurlod · Post by **ray.wurlod** » Mon Mar 03, 2008 3:05 pm

I couldn't have put that any better. There is also a mindset difference; with Informatica one tends to design from the bottom up (take the data layouts and build an ETL structure on those), with DataStage one tends to design from the top down (draw the "big picture" and fill in the details).

lstsaur · Post by **lstsaur** » Mon Mar 03, 2008 3:34 pm

Robert,
As long as you are willing to train your new employee on DataStage, you should just make sure that person has "GOOD" SQL and Unix skills.
With one week trainning on DataStage, the new employee should be able to start to design the jobs. Why worry about how difficult to transition from Informatica to DataStage.

DSguru2B · Post by **DSguru2B** » Mon Mar 03, 2008 3:36 pm

...and basic ETL understanding, conceptually.

chulett · Post by **chulett** » Mon Mar 03, 2008 3:50 pm

lstsaur wrote:Why worry about how difficult to transition from Informatica to DataStage.

Because not all people will be able to make the transition. Sure, it really shouldn't matter but you'll find people still wanting to shove their square peg in your lovely round hole, complaining all the time that it ain't the square they are used to.

ray.wurlod · Post by **ray.wurlod** » Mon Mar 03, 2008 5:14 pm

Don't need Informatica in the equation for that. Hence the need/demand for a Server to Parallel Transition Lab to help folks to unlearn some server job "square pegs".

rhale · Post by **rhale** » Mon Mar 03, 2008 6:10 pm

So let me summarize and see if people agree or disagree.

While general ETL, SQL and UNIX skills will transfer, specific Informatica skills may not and could be a hinderance. Therefore I should focus on candidates with strong general ETL, SQL and UNIX skills with the knowledge that I am going to have to train specific Datastage skills.

ray.wurlod · Post by **ray.wurlod** » Mon Mar 03, 2008 6:25 pm

You've got it. Look also for traits such as attention to detail, accuracy, initiative and the ability to complete tasks with minimum supervision (but willingness to seek clarification in cases of uncertainty rather than ploughing ahead).

Virupakshann · Post by **Virupakshann** » Tue Mar 18, 2008 3:32 am

100% correct

vipson · Post by **vipson** » Thu Mar 20, 2008 9:32 pm

With my own experience, I say you are on the right track.
Just make sure they have good ETL skills and enough SDLC experience...DataStage can be taught and they should be picking up fast..Ask them not compare Informatica all the time. They may not like the DataStage documentations and resources..You have to be patient with them.

eostic · Post by **eostic** » Thu Mar 20, 2008 10:02 pm

Ken...

Thanks! Your points were very interesting, and I've heard similar things, but from a different point of view --- I've had numerous folks experienced with that other tool who have said that the needs arise in the same way (for example....[fake example for illustration] ...."gee, I need to code a Sleep routine that takes two columns, concatenates them and then counts the characters, and then Sleeps for that many milliseconds")...and in DS, even the "average" developer can code that up within the tool, with very little programming skills, while the other tool's developer has no choice but to go "outside" and do it in <name the language>, requiring a compiler and IDE, etc.

EE changes that game a bit, of course, but even there I can do some things "in the tool".

Just an observation --- anyone know how true that still is?

Ernie

kcbland · Post by **kcbland** » Sun Mar 23, 2008 9:50 pm

Just a little more feedback. I just came back from a consulting week with 16 very savvy people, several of which are accomplished Informatica folks. Each represented a different group from a large global company looking.

They were stunned with DataStage. The simple Server side Function test harness blew them away. The fact the test cases stay with the function forever was simply amazing to them. The scalability, the functionality across Server and PX, they looked dazed (in a good way). This stuff sells itself to programmers.

Folks kept saying things like "if I only had this when we did xxxx project". To say I had an awesome week would be an understatement. I felt kind of like a chocolates salesman introducing my product to folks who had never had chocolate before. I couldn't show them enough in a week all of the things we could do in DS.

I think the hard part in DS is picking the method to use in a given situation because there's so many ways to do things.

jdmiceli · Post by **jdmiceli** » Mon Mar 24, 2008 2:26 pm

As someone who spent a couple of years working with Informatica before joining the DataStage world (Server), I can attest to the difference in mindset that Ken, Ray and others have mentioned above. The two products are comparable at the basest level in defining their ultimate purpose: to move data from heterogenous sources to an ultimate target. I thought it was very left-to-right and the mappings were consistently very large.

I agree that Informatica is more of a mapping level tool (not that that is a bad thing), but the drag and drop mentality with little ability to tailor the operation of the individual objects is pretty much built in to the product. I found that to be somewhat limiting with the version I was using at the time (5.1 - it has been a few years). Whereas DataStage uses what I have always called an 'object' based methodology which uses small jobs to perform a particular purpose and several to many jobs like this are strung together to perform the ETL processing. There is a great amount of flexibility because we do have the ability to alter the way the jobs work using code, in the form of functions, external code (scripting languages, shell stuff, file access at the OS level, etc.) and the like.

It is definitely a different paradigm that took some getting used to. At first, I hated it; but now that I have gotten past much of the initial learning curve (not that I'm any expert yet), I find that I favor DataStage over Informatica as an enterprise level ETL tool.

I would suggest that you look for a solid understanding of the theory behind ETL and database design and processing in general, as well as Unix/Windows skills. With those skillsets solid, the semantics of the tool of choice can be taught/learned with relative ease.

Hope that helps!

Sudhindra_ps · Post by **Sudhindra_ps** » Mon Mar 24, 2008 3:13 pm

hi,

If one has good amount of experince in Informatica, he could apply the same concepts in Datastage too. As Datastage has various stages(Transformations) in it which can support any complex logic to be applied. The basic criteria to work with any ETL tool, one need to have solid understanding of OS knowledge,SQL skills and good programming techiniques. Having these qualities any one can feel the ETL tool with which he/she is working is the best one to work with.

Thanks & regards
Sudhindra P S