Since march of this year (2019) our customer is experiencing extensive performance loss due to the AMP process.
We have noticed that at the beginning of a WCF server call the AMP service (epsecurityservice.exe) becomes very active (50-80% CPU) for quite a long time (30-90 seconds). We don't receive an answer from the WCF-service during that time.
When we shutdown the virusscanner on the client we can see some improvements (20-30 seconds). Shutting down the virusscanner on both the client and server will give us the expected response time (less than 1 second).
Excluding the server and client process will help some, but still doesn't give us the expected response.
The WCF service is using a custom Binding with binaryMessageEncoding and compression using GZip on net.tcp (port 8733).
Our questions:
- What is causing this, and why didn't it cause any problems before?
- What is the preferred solution?