How to measure job complexity?

WoMaWil · Post by **WoMaWil** » Fri Jun 17, 2005 1:18 am

I think, that all of the top 25 posters in this forum are able to judge with their eyes and fuzzy experience DataStage jobs ex-post. I've never done so to set up a list but in my mind I could imagine, that there may be a list of some dozends of core data constalation and there is an optimal aproach for each of this problems we could write in a handbook Case 1 to Case n.

Sometimes you have a simple job and nice to understand job, but you are not happy with the time it needs, so you make this job more and more complicate to make it run quicker.

From the list of the large amount of the help posters everybody might be able to rate a DS developer and her/his skill, but to find objective criterias or test will not be easy.

Wolfgang

lshort · Post by **lshort** » Fri Jun 17, 2005 12:03 pm

Im not sure I want to measure in this way. A lot of Datastage (may I dare set..."at our level") is akin to art.

Can you measure the worth of the works of 'da Vinci' or 'Mozart' by number of brush strokes and colors used or by counting the notes.

kduke · Post by **kduke** » Fri Jun 17, 2005 1:11 pm

If these guys were paid by the hour then yes I would try to figure out who did the most work. As a business man then I am required to find good talent and make sure they are productive. You also want to make sure that what they create is good quality and easily to modify. Those that think it is art never had to fire one of these guys. I agree though when done well it displays a beautiful mindset. Can most companies afford art? If so why is the lowest bidder building these data warehouses.

lshort · Post by **lshort** » Fri Jun 17, 2005 3:09 pm

My issue with the clinical quantative approach is that is does not take into account other important variables like ...

How well the spec is written?
How good is the developer at interpreting the spec?
What level of rapor has the developer cultivated with the client
-ability get the requirements in an efficient an timely manner
-ability to get the cooperation of DBAs, Unix Team...etc.

Each of these things and more go into determining what it ultimatly takes to get the job done well on time and within budget.

This becomes painfully obvious to the client that goes for the lowest bid.

kduke · Post by **kduke** » Fri Jun 17, 2005 4:18 pm

Well said.

ray.wurlod · Post by **ray.wurlod** » Fri Jun 17, 2005 5:47 pm

One of my main differentiators is high quality documentation of everything I do at the customer's site, another is that I try always to mentor at least one person in my techniques so that anything I leave can be maintained. Elegantly simple designs aid in this.

One of my main success criteria is that I have an almost 100% record of getting repeat business.

vinodlakshmanan · Post by **vinodlakshmanan** » Fri Jun 17, 2005 7:09 pm

I agree totally. And one of the things we often need (since we have many small modules) is to give an estimate of the complexity of the jobs.
How do we determine what is a simple jb, a complex job or a mixture of both?
I may be doing only 1 lookup in my job; but (as in our case) there is no metadata importing tool as we get data in files, so we "type" in the fields. But if it has 100 fields should it qualify as complex?? I do not think so.

Maybe we can measure efforts as a combination of all the following cases:
Design effort - actual logic put in, simplicity, basically "IQ" involved. Would involve rework due to performance issues, logic faults, etc.
Coding effort - this would involve (in our case) "typing in the metadata". This would not be ne involving IQ
Testing effort - testing for correctness of design, testing for live data, performance testing

WoMaWil · Post by **WoMaWil** » Sat Jun 18, 2005 3:27 am

Hi Kim,

the questions you ask, may be asked, but are not easy or even impossible to answer. You have to value things. A ETL-Job done in PL/SQL may perform in less time and even you may need less time to write it, but is it per se better?

Ray is leader in in this forum quality and quantity, but may be there is somebody who does it the same but we don't know him. And perhaps there are some problems somebody else had a better clue.

So Kim, you can approach a ranking and a messuaring but you will never find the final path.

I once worked in a company for furniture after sales service with 200 field technician. The boss had a fine system how to count their work. But in less then 6 week the technician had optimized their appoach to max their income and the boss had to start to change his meassuring system.

Wolfgang

chulett · Post by **chulett** » Sat Jun 18, 2005 6:43 am

:D Something similar going on at the company I'm currently working for. Salespeople are tracked and compensated by two different systems. The tracking system is very cool and quite flexible when it comes to dealing with all of the tricks people have come up with to 'beat the system'.

On the compensation side, however, there are two major classes of work: a) the kind of sales that benefit the company long term and b) the kind of sales that benefit the salesperson (and their bank account) short term. Unfortunately, at the moment, they are not the same sales. Given a choice, can you guess which one they will put their effort into?

Not sure what my ultimate point is, other than to say that no matter the yardstick by which people are measured human tendancy is to do whatever compensates them best on that yardstick, not whatever is best.

Ferengi Rule of Acquisition #13 - Anything worth doing is worth doing for money.

kduke · Post by **kduke** » Sat Jun 18, 2005 8:29 am

Craig, another great point. This split into so many different directions but all very related to comlexity and the measure of work done.

Next time anyone asks you do describe the complexity of your work then print this topic. It should help.

ray.wurlod · Post by **ray.wurlod** » Sat Jun 18, 2005 6:05 pm

Or just cite the DSXchange Complexity Scale, which ranks complexity on a scale from 0 to 50. And the answer is (naturally) 42.

ray.wurlod · Post by **ray.wurlod** » Sat Jun 18, 2005 6:05 pm

Adds a whole new meaning to DCS, doesn't it?!!

Ya hafta have a three letter algorithm otherwise IBM doesn't understand what you're talking about.

roy · Post by **roy** » Mon Aug 15, 2005 8:46 am

Hi,
I guess I missed this one

,
My question is who will pick up the task of making a decent FAQ post from this

Anyone?

By the way, IMHO, this is a perferct FAQ discussion forum post that should mature to a FAQ post
I would like to get approval for moving this post over there!
KCR (3 leters algorythm that stands for Kim, Craig, Ray

) Do you all agree?

(for those wondering I decided to go 15 pages back and go over all posts, so i came across this post; kinda prooves my point in my recent post at the editor's forum regarding unread posts)

kduke · Post by **kduke** » Mon Aug 15, 2005 10:06 am

I am not sure I want my rants to be so easily found but go for it if CR agree.

roy · Post by **roy** » Tue Aug 16, 2005 12:51 am

I Think that at least the job complexity if not more stuf in this post could make a good FAQ Discussion post at least.
Don't you?