Qlick view freezes when loading multiple sas data sets.

[#1]

Hi every one, I'm new to Qlick view and this forum, so i apologise if i have posted this in the wrong area. I'm attempting to generate some reports from some SAS data files i have stored on my machine. I appreciate SAS is probably not the most common data source so there may not be a wealth of experience present on this in the community. I have set up ODBC connections to my data sources and can import individual data sets no problem generate tables etc..

However the problem occurs when i try to access multiple data sets, Qlick view loads the data fine however after the last data set has been read, it just freezes and the elapsed time remains stuck.

I'm also stuck on what to do any suggestions would be great.

Connecting to Shr1
Connected
TEMP 115,008 lines fetched
Connecting to Shr1
Connected
TEMP 106,727 lines fetched 

[#2]

Hello Mark,

I am not a SAS expert, so I can't tell you any thing about possible lost connections, time outs, buffer overflows or similar things that may happen.

But as you said, "after the last data set has been read, it just freezes and the elapsed time remains stuck." here a guess:

After loading the last data set QV links together the internal tables via same fieldnames. May be you are reading a lot a fields with same names (case sensitive) and QV is struggling while trying to link all the tables together ? Take a look a the Task Manager --> CPU and RAM usage of your QV-Process; still alive ?

HtH

Roland

[#3]

Hi Roland,

Thanks for your reponse on this. I am sure the error does not fall on the ODBC connection as i have tested the link within MS Access, I can also monitor the sate of the link within SAS, and there are no errors.

However your point "After loading the last data set QV links together the internal tables via same fieldnames" Is something i hadn't considered, due to the nature of my data there are a lot of identical field names, which could be a problem i guess...

I have since tried this load on much smaller data sets with far fewer field, and success it worked. However this is not really a complete solution, as in future i will need to evaluate multiple SAS data sets with numerous shared fields.

I will proceed to check the state of the process on these larger load operations to see what is happening.

Thanks for your advice

[#4]

Hi Mark,

glad to help you (a bit).

Please keep in mind the automatically linking of tables with same fieldname(s). This is a very useful feature to ease data modelling during the load. But if you are not familar with it, it may lead into unexpected, funny results (like so called sync tables and(or) even to tables being linked without any sense). To check the QV data modell you can use the Table Viewer (<ctrl>T or smart icon).

Regards, Roland

[#5]

Hi Roland

Glad there are people like you in this community Roland .

Yes i had noticed some of the joins it created seemed a bit strange. Is there any easy way to turn this feature off that you are aware off? - It not a feature i will really need.

Many thanks again.

Mark

[#6]

Hi Mark,

no problem.

Check out the (un-)qualify-statement. With them you can turn off/on the use of fully qualified fieldnames with a preceding tablename like MyTable1.FieldnameXY.

In the first step all fields should be fully qualified. The result is that all tables are "orphans".

Next step(s) is to identify tables which belong together and should be linked together using the unqalify "keyfieldnames". May be you can even use QV-Joins, but this (at last) should be a topic of a training lesson.

HtH once more

Roland

[#7]

Another lengthy procedure, is to identify the field names that are linked (same field names from 2 or more tables), and manually aliase them (eg. Field1 as Temp_Field1), to break the association created by Qlikview.

This is helpful when there are less associations, automatically created by Qlikview. And this also helps you to name a field with a proper naming convention.. But if there are too many associations which is not required, then the method suggested by Roland is the best way.

Thanks,

Robinson