Embedded Capacity size required for refreshes with Powerbi Pro

[chrisnewman]

I have scheduled refreshes set up for 8 datasets, each around 350Mb. 2 starting at 5am, 2 at 5.30am, 2 at 6am and 2 at 6.30am.  They are refreshed through a gateway. Nowone else is expected to be using the embedded at this time. What size capacities should I be using because they keep taking too long and failing.  I've even set up to be A3 capacity now and still having problems. I wouldn't have thought these should take that long. Any thoughts or whether I'm doing something wrong much appreciated.  

[colivier]

Hi @chrisnewman 

Do your reports take significantly longer to refresh when published compared to when you do a refresh using Power BI Desktop? If so, the issue may not be with the size of your capacity but with other factors such as resource constraints on your gateway, configuration of drivers on the gateway server (if you use 3rd party drivers), network, or the speed of the actual data source, e.g. SQL Server.

Refer to the following article on how to monitor and optimize the performance of your on-premise gateway: https://docs.microsoft.com/en-us/data-integration/gateway/service-gateway-performance

You can also try temporarily deploying an additional gateway node to spread the load of the refreshes. If it helps it may be cheaper than upgrading your capacity. 

Also try analyzing your reports with PBI Spy to check if all the data columns that are imported are actually used in your reports. Keeping your reports lean will result in faster refreshes, less memory usage in the capacity, faster response, etc.

You can download PBI Spy from: www.pbispy.com

Feel free to contact me if you need any assistance

[v-janeyg-msft]

Hi, @chrisnewman 

It’s my pleasure to answer for you.

According to your description,I think you can try to advance the refresh time and increase the interval, because the refresh time you set may not be on time, there may be a 15-minute delay.Secondly, it isn't recommended that many people use embedded capacity when refreshing, which will cause resource shortages.

If it still doesn’t work, you may have to increase the capacity like a4.However,However, i think you can use 'refresh now' for each dataset on the service, and  check the refresh history on the setting of the dataset ,then detetmine new schedule refresh time and interval according to different refresh time, then set a reasonable schedule refresh plan.

If it doesn’t solve your problem, please feel free to ask me.

Best Regards

Janey Guo

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.