QVW size vs RAM size

[#1]

Hello everybody,

I want to know how qvw is represented in RAM when loaded into Qlikview Desktop client.

For example: I have a small sample application of 300KB but when I load into my desktop client, QV.exe process size increases by 9 MB. So basically 0.3 MB file expands to 9 MB when loaded. Can anybody explain this?

thanks!

Mayank

[Gysbert_Wassenaar]

.qvw file can be compressed. When the file is loaded in memory it is decompressed. You can check the level of compression on the General tab of the Document Properties window.

[sudeepkm]

According to QlikView Technical Brief System Resource Usage document.

The size of document in memory would range from 2X to 10 X of the original size (or on disk size)

With additional concurrent users the size grows 10% of the initial size in memory per user.

[#4]

True. But here increase is around 30 times which is wondering me.

Also, I had a 333 MB QVW file. It has a Straight Table with conditional calculation and around 0.3 Mn records. When this condition is disabled and records are allowed to show up, RAM in Task Manager increases from 30 GB to 70 GB and finally to the top limit after Allocated memory Exceeded message.

Questions:

1)How can a 333 MB qvw file can go beyond 40 GB in RAM and fails to load when calculation condition is disabled for the straight table?

2) Again, RAM is increasing over 100 times the file size. How come?

Mayank

[sudeepkm]

what I mentioned is as per QlikView Tech document and applied to ideal qvw case.

If a qvw has been built using a resource heavy logic then the RAM size will go beyond 10 times.

for example I had a qvw disk size of 14 MB which had 100 K records only. I had a pivot table which had all the records at detailed level with multi level If else conditions.

And when we open the QVW it took around 8 GB at each node level of the Pivot table.

I had to optimize the qvw by avoiding multi level If else and implementing set analysis.

So you may have to optimize the QVW.

[jonathandienst]

Hi

If you have a chart/table where the data comes from 2 or more tables from your data model, and the tables are not properly associated, the QV will be making a cartesian product of the data tables to perform the chart calculations. This can take up a lot of RAM. For example, 2 tables of 100K will result in a cartesian product of 10 billion rows, and 3 tables of 10K each will result in a cartesian product of 1 trillion rows (and an out of memory error!). That's how a small qvw could require huge amounts of memory.

HTH

Jonathan

[#7]

Size of document in local x number of users x 4.

This is from a class i took in past. I think it still applies.