如何使用Azure通过Web服务加快数据性能的插入

Question

I am trying to insert 6000 rows into a database. It takes around 1 minute for that operation.

My architecture is as follow:

Console app -> Web service with endpoints -> sql database

Console app is running as a webjob on Azure, and is using Premium 3(P3) plan with 1 instance scale out. Console app is using Parallel.For for faster insertion.

Web service endpoint is using P3 plan with 8 instances. Amount of instances from 8+ doesn't make any change, so problem is somewhere else. When I am checking live telemetry I can see there are 200 requests per seconds, and request duration is 200ms average which means web service is not bottle neck.

Database is P4 and DTU percentage is below 30% so that is not an issue as well.

One thing I found is that, when I run Console App from localhost computer I can finish operation in 50 seconds, in comparison to azure webjob which takes 1 minute. I tried to scale webjob to 4 or 8 instances with no luck.

What could I observe further to find out if I can finish this operation faster with Azure?

Answer 1

I tried to scale webjob to 4 or 8 instances with no luck.

There are two types of WebJob. Continuous and Triggered. Only Continuous WebJob can be scaled out and run on multi instances. You could choose the WebJob type and Scale type when you creating the WebJob.

You also need to rewrite your code to suit the scaled WebJob. For example, if you want to insert 6000 rows to a table, you could insert 1500 rows by each job if the instance count is 4.