After migrating one of our websites that uses Umbraco CMS (v6.2.6) from on-premise web and database servers into Azure, there have been huge performance implications for users in the back-office. I have tried scaling up the resources but this is not the issue as the DTU utilization only maxes out at 35%, even when bursting / spiking is seen. The SQL server is S0 standard service tier (10 DTU's) and the resource group that it sits in is S3 standard service tier (4 core / 7gb) - located US East.
I duplicated the azure app service and sql database and attached an instance of New Relic to it in Azure and the findings are a bit disturbing.
If it makes any difference, I ran the same code / tests locally with a local instance of SQL server and the response times were much quicker - with little to no time being used to open the connection - Localhost StackifyPrefix Trace. I didn't write the data-layer code so adjusting it at this point is definitely out of the question, as cringe-worthy as it may seem...is there anything else that can be done to mitigate this? That seems like an absurdly long time to just open the SQL connection a bunch of times