Why is apps?chacheburst a.k.a api/hub/ so slow? a.k.a why is opening the HUB so slow?
The response in our environment for this API has 181,034 characters for me when opening the Hub. It sits in Waiting (TTFB) for 7+ sec for 280 apps. And it doesn't change if you open the Hub on a Stream first...
Qlikview Accesspoint opens in half the time. Is Qlik Sense a big step backwards in Hub performance? Or are we doing something wrong?
Re: Why is apps?chacheburst a.k.a api/hub/ so slow? a.k.a why is opening the HUB so slow?
That endpoint (/api/hub/v1/apps) is the resultant of the Repository's evaluation of what apps a user can see. I'd expect initial load times to be higher than subsequent load times, as the Repository does some level of caching.
As you mentioned, the response does not change between opening up /hub and opening up a specific stream.
In terms of the cause, it's hard to get into specifics for a general problem but it is the interaction of:
The number of total apps in the environment
The amount of meta-data that the Repository has for the user (e.g. AD groups)
The number and complexity of the security rules
As for the comparison to QlikView, at least some level of this is due to the sophistication of the rules engine in Qlik Sense. In QlikView, you have NTFS or DMS.
NTFS does an ACL lookup on a file and compares against the user
DMS does a lookup against the .meta file
In comparison to Qlik Sense, where the default behavior (determined by the Stream rule) uses inheritance from multiple levels (e.g. Stream > App).
I'll reach out to you privately to be a beta-tester on a tool + an app which may be able to help.