How to handle large QVD Size

[agni_gold]

Hi all ,

How can i manage my qvd size , if my qvd is of 5 GB so it will take time to load and when it loaded for my document then it will make my dashboard also very big , how can i manage these things ....

[petter]

Reduce the number of fields to exactly what you need. Try to optimize each field so you get a highest possible cardinality - ratio between unique/distinct values and total values for each field. That will improved the de-duplication that QlikView use to minimize storage and maximize performance.

If you have a high number of fields you might actually split the table vertically and use different vertical subsets for different dashboards. You could also partition/split it horisontally usually along natural time intervals like years, quarters, months and weeks.

Then you shoyld also consider whether it is possible to split your dashboard into several dashboards.

[agni_gold]

I can do the spliting of qvd but the problem is that , user wants all data in the dashboard for all years , so if i split then i have to load all data in dashboard .

So have u any idea about this ...

[petter]

Splitting the dashboard into multiple can be done seamlessly if you do it in a smart way. Achieving your necessary optimisation and keeping your customers happy. You can use buttons/text boxes to look like tabs and use an Action to open another dashboard that looks like the same dashboard - and transferring state too so the selections are carried over.

The optimisation of fields are still a valid and could give you huge space savings. In one media company that had an analysis app for set top boxes I was able to squeeze down the size of the app footprint to 1/8 of the original size by just truncating a DateTime field to the time resolution necessary - it had been running for months without this optimization.

[agni_gold]

So what is exactly partitioning , why we use this 

[petter]

You can decouple the UI from the data model and have multiple parts of the UI that seems like one app to be split in several apps that connect to several partitioned/split parts of the data model.

It is just a technique to handle larger applications with larger data models so they still perform adequatly.

Whether that can work for you or not depends on a number of factors. And you have to do a careful analysis to be sure it will work.