Set Analysis and RAM Performance & PC Hanging

[#1]

Hello all,

Just a quick question performance-wise.

I have a request from client to produce a spreadsheet with 18 yearly tables all filtering the data in different ways. I have a QV app with 18 bookmarks and then a copy/paste process of tables into Excel. This is not optimal and since the project and development of ad-hoc tables need to continue after my departure, I had initially chosen to try and implement 18 objects using set analysis with bookmarks (thinking it would be easier to hand-over with minimal training) and a macro doing the export into one spreadsheet.

However, this 40Mb document hangs the PC almost systematically, as when I create the set analysis charts (by about the 10th), the Task Manager shows this 40 Mb file with QV open take up 3 Gb of RAM. Initially I thought it might be the bookmark set analysis, so I explicitly stated the set analysis and... same result.

Has anyone seen this kind of QlikView "desktop" behaviour? Any explanation as to why it does this? Any workaround (apart from throwiong more RAM at the machine)?

Kind regards, Cheenu

[iliyansomlev]

We have also experienced the same issue recently. A 20 MB document opened locally on the server machine with Desktop client on a 1 GB RAM server takes 500 MB RAM and dashboards and reports work fine with good speed of calculation. When opened in 4 GB-RAM server it takes 2 GB of RAM and opens and calculates the same speed. When opened in 16 GB-RAM server it takes 12 GB RAM and has same performance and when opened in 512 GB- RAM server it takes 71 GB of RAM !!

We have a normal (snowflake-like) model with 3 very small island tables, the biggest table has 2 million lines, and dashboard and other object use set analysis.

Does anyone know what is causing this strange behavior of QV desktop app?

[Anonymous]

Same issue here, and it's the same result on the server! (with 128GB RAM)

The SQL query would be as simple as:

SELECT COUNT(A)

FROM TABLE

WHERE B > 2*C

GROUP BY D

Table would be like:

A     B     C      D

1     10     30    A

2     20     25    B

With a few million lines.

It seems Qlikview can't be used to do some trivial aggregation when you have to compare two columns of the same table!