6 Tips to Get More Out of Your Capacity-Based License

In our previous blog we described the new licensing model for Qlik Cloud Analytics. In this new model, the key metric is the amount of data being analyzed in your platform. All the data you load or create in the platform, in bytes, is added up and measured against the capacity you've purchased. This means you're directly rewarded for setting up your Qlik environment smartly. In this blog, senior Business Intelligence consultant Lennaert van den Brink gives you 6 tips on how to use your Qlik environment as efficiently as possible. Bonus: even if you don't have a capacity-based license, many of these tips will help optimize the performance of your environment!

#1 Optimize your data models

We often see that a large portion of the columns we load aren't actually used. That's a shame, because every bit still counts. You can also often gain a lot by, for example, using autonumber for key fields. Many of the possible optimizations are already useful anyway because they improve your dashboard's performance, but in the new model you'll also see the benefit in hard euros.

#2 Think carefully about how you set up your Qlik Cloud tenant

In the old model you weren't charged for duplicating files. It was therefore sometimes useful to, for example, write QVDs to multiple spaces for self-service use.

#3 Clean up your environment regularly

A dashboard or QVD that's no longer used but is still refreshed regularly counts toward your total, so get rid of it!

#4 Consider using parquet files instead of QVDs

It's recently become possible to read and write parquet files directly from a Qlik script. Our internal tests at E-mergo show that in some cases the resulting parquet files are significantly smaller in file size, especially when the dataset is sorted before being written. The downside is that these parquet files can sometimes be slower to load than QVDs (optimized load). So it's worth weighing up whether you prefer faster processing or lower consumption.

#5 Think carefully about your large apps

Of course, we'd all prefer to always load in all the data, but do we really need the detail level of invoice lines for 10-year-old invoices? You can choose to load older data at a higher aggregation level, or perhaps not load it at all. Do you still need the detail level? Then you might be able to use on-demand app generation or direct query so that you still don't have to load everything into the environment.

#6 Consider pushing transformations back to the source system

If your source is a transactional system, such as a point-of-sale system, this may not be so practical. But if you're loading from a data warehouse, for example, there's a difference between loading 3 tables with SELECT * and then joining them together in Qlik with JOIN statements and WHERE filters, versus putting this directly in the query to the source system. Of course, you could also ask your data warehouse developers to implement these transformations on their end.

Written by Lennaert van den Brink
Senior Consultant