Unlock a world of possibilities! Login now and discover the exclusive benefits awaiting you.
Hi,
i have the requirement like i can use join before loading the DB data into Qlikview or after loading the data from DB(sql) can also perform join
suppose left join on two tables.
Which one is better performance.Is it better to put joining burden on DB or to put the joining burden on Qlikview.
Which one is better option Performance wise give me scenarios.
Regards
Mahesh
It would hepl, if you describe the amounts of data (rows, columns), but overall DB side is better...
I have done qlikview joins also with quite big # of rows (>700M), but if it is possible, then i prefer order some views into the DB from IT guys...
It would hepl, if you describe the amounts of data (rows, columns), but overall DB side is better...
I have done qlikview joins also with quite big # of rows (>700M), but if it is possible, then i prefer order some views into the DB from IT guys...
Hello Mahesh,
I believe this decision will come after a little homework. After all, there are too many unknown factors at this point that need to be considered. In fact, it depends on the type of hardware your source database is installed on, the type of RDBMS you are using, the amount of rows that are to be loaded, the amount of memory available to QlikView, the period of the day the load takes place, etc.
For small tables, I don't foresee any difference in performance, but for very large tables, it might. Also, depending on the type of join, in one case it can be very heavy for a database to manage, while in QlikView, with a few ApplyMap() to replace codes with descriptions, for example, it does not affect the load process very much.
To be known, in RDBMS's such as Oracle, the engine can precompile statements and keep them in cache if they are used quite often. So for complex joins, maybe you can create a view in the source system?
In brief, while there aren't any pre-defined answers for your question, there might be a lot of possible solutions, all depending on your current implementation.
Hope this helps, regards,
Philippe