Is Power BI report server naturally slow?

[opejoseph]

So i recently moved my power bi assets from Power BI service to my self hosted report server. I noticed it seems loading a report loads slowly when compared to power bi service. I am using a good remote server that has some capacities. I have done the best practices like reducing the joins in the report, cut schedule refresh, and optimizing queries, using the performance analyzer on desktop shows positive result too.   Stakeholders get pissed off when it takes forever to see anything on their screen. Can anything be done to mitigate this?

[mpsrshl]

 It run faster in our local server than in the closest datacentre in our tests. 

[我是天才啊11]

I used a very simple report for testing, which can be opened in the cloud in about 3 seconds, but it takes 5 seconds locally

[josef78]

First, analyse ExecutionLog3 to check what is slow, if model processing, open report definion, model loading to memory, or dax queries. Second, check you server resources, especially memory a CPU utilisation. Also check how often are report wipe-out from memory.

For comparsion, in my enviroment, I must install second PBIRS server in scale-out mode, and dedicate background model processing to second server (and first server only as PBIRS front-end). My experience, model processing, and also dax query performance is fully comparable to cloud Power BI Service. But report open (authentificate and definition load) and load model into memory (if not already loaded), is little slow (but not big difference}

[我是天才啊11]

and dedicate background model processing to second server (and first server only as PBIRS front-end) Can you explain how to do it？

[josef78]

On Front end servers, find rsreportserver.config file in your instalation folder. And set <IsDataModelRefreshService> to false.

[我是天才啊11]

thank you

[josef78]

First, analyse ExecutionLog3 to check what is slow, if model processing, open report definion, model loading to memory, or dax queries. Second, check you server resources, especially memory a CPU utilisation. Also check how often are report wipe-out from memory.

For comparsion, in my enviroment, I must install second PBIRS server in scale-out mode, and dedicate background model processing to second server (and first server only as PBIRS front-end). My experience, model processing, and also dax query performance is fully comparable to cloud Power BI Service. But report open (authentificate and definition load) and load model into memory (if not already loaded), is little slow (but not big difference}

[lbendlin]

https://www.gigxp.com/powerbi-report-server-performance-tuning/