Breaking out timestamp field to decrease cardinality?

[Sazabi]

Hey Qlik Experts!!

I have a question on app optimization:

I've read that a good way to optimize apps is to remove or break out timestamp fields to increase performance.

I read somewhere that if timestamps are necessary, it is best to split the timestamp field into Date and Time fields.

From an app consumer perspective, how are they supposed to properly graph on a line graph timeseries data when the Date and Time field are broken apart? 

Dimension 1 would be Date...but how would they graph the hourly portion? Field concatenation? Wouldn't that be slower than just having a regular timetamp field?

Hope that made sense; I can clarify further if needed.

Thanks!!

S

[Vegar · Accepted solution]

The reason why you get this recommendation in Qlik Sense is that it store each unique value once. If you don't split then almost every value will be unique. If you split then you will only store one date value per day and the total amount of time values are limited.

The recommendation is to consider splitting timestamps, it sounds to me that you need the timestamp field in order to create good visualizations for the end users. You can however consider if the timestamps are to detailed. Are they stored down to milliseconds or just minute? What are your needs, do you need milliseconds, seconds or minutes in your visualization? If not then consider rounding the timestamp to you desired need. You might have 10 unique timestamps within a minute, so if you round them down to the nearest minute you will only gave one unique value for the ten transactions. 

[Sazabi]

Ah I understand.

In my case, I have hourly data which is the required granularity, meaning it's probably at the lowest it should be.

Thanks for the clarification!