denzilsyb wrote:
the code is great, but if you develop the code and then leave the poor client with BASIC code that is complicated to the untrained eye, you arent doing them any favours.
We will have to agree to disagree on that one. I'm absolutely for KISS. We've got one routine to run jobs, a couple of utility function to create/delete trigger files and that's it. The scheduling jobs just consist of a sequence of "run job A", "run job B", it doesn't get any simpler than that. You want to add a new job: copy and paste 5 lines of code and change the name of the job to execute.
denzilsyb wrote:
Ive seen a pretty complicated sequencer a colleague of mine put together; and it worked. I guess its just a matter of changing how we die hards think about the sequencer.
I've seen sequencers that were also very complicated, and then one upgrade they just didn't work anymore as they should.
Some of the things we now do couldn't be done with sequencers anyway (or would make the sequencers enormously complex).
denzilsyb wrote:
You say that you let the DB do just about everything. Why do you think DS is performing so bad on your side (except for aggregator and db2 load)?
Never said performance is bad. It isn't because we do most of our stuff on the database side. If e.g. you select 1.000.000 rows from the database and use a simple constraint in DataStage to select 50.000 rows, or you just select the 50.000 rows already in the database, I can assure you the last version is faster
Aggregating in DataStage or database... don't have to think 2 seconds on it.
The way I write my jobs I most of the times I don't even use lookups or hashfiles... I sandbox my dimensions with business key/surrogate id in the staging area and just "left outer join" away in the SQL input query... 90% of the time faster than using a hashfile/lookup. So you could say I use DataStage as a glorified data mover.
Besides the major databases Oracle/DB2/SQL-server have a lot more customers than Ascential so as side-effect they contain a lot less bugs/quirks (before I install them anyway
)
I did grow kind of used to DataStage and its quirks, but if tomorrow someone wants me to implement a datamart/datawarehouse in Informatica or DecisionBase I won't lose sleep over not using DataStage. DataStage is just a tool, the important thing some developers seem to forget is the overall design of your ETL.
My 2 cents
Ogmios