Qlikview to Qlik Sense migration as an opportunity to improve?
My client is starting a migration from Qlikview from Qlik Sense, and even if there is no rush, business units and managers are convinced by the new product and wants us to move forward.
At the moment, Qlik is the only consumer of data for reporting and it should stay like this in the next 2-3 years at least.
My client's guidelines for Qlikview development are the following:
Use only SQL queries which we run in qlikview scripts in order to join source tables and do some transformations (The reporting data structure is therfore very similar to operationnal system)
No staging area in order to store raw data from source system
We should not use Qlik platform to historize our data. But we don't have anywhere else to put data: we don't historize. If there is a need to historize (change in dimension for example), we backup the front-end application...
Qlik models are based on operationnal systems and no conformed dimensons/facts are in place
This situation is mainly like this because the guy in charge keeps saying that Qlik Sense and Qlikview are not ETL tools, that we have to keep it easy for us, that BI team should not know data structure of all source systems and that we should wait until an ETL with DW comes.
It's been like that for years and even if management begins to understand that data should be better handled, situation will not change in the following months and there is no guarantees that it will one day. We also know that most of our source systems will be replaced in the next 5 years.
So I suggested to take the migration toward Qlik Sense as an opportunity to "improve" the way we manage data in our Qlik platform by
Using ETL capabilities of Qlik
Setting 3-Tier layer architecture (1-Extraction of raw data/2-Transformations/3-modelization and presentation layer)
Creating new datamodels, that should not be based on existing operationnal system (use Dimensional Modeling). Existing model should be reviewed and maybe we should start from scratch if the gap between existing model in Qlikview and target model is too big.
After many heated meetings, this approach has been rejected by the guy in charge, and we are likely going the "quick and dirty" path.
I have nearly ten years of consulting in Qlik technologies but today I am seriously beginning to doubt of myself, could you tell me if my approach is really that crazy?
Let me tell you that, it's interesting to read your post. I have little less experience (9 years) than yours in qlik. And today, I believe that this is something about non-technical skill, and that is related to convincing people going in their way. I would rather try to convince 'the guy' by showing benefits if qlik is used more like it should be. I would also try to remove his reservations by explaining the ways of mitigating the risks he might be seeing with qlik scaling.
I totally agree with that, but it is the first time that I see so much arguing about a this topic.
It all started because they asked an assessment of their BI platform. When we presented our findings and recommendations, it triggered an epidermic reaction at him, even so we were not pointing anybody and saying that someone had done something wrong. We have even been asked to softened the findings so that management does not think that BI platform was in a critical situation 🙂
We have a lot of PowerPoint slides based on official Qlik document, best practices and previous projects showing benefits of this approach and how risks can be mitigated. But after nearly 4 months of spreading the good word, I am thinking that the cause is lost.