Summary

In a shared environment, CPU, memory and network resources used by each container should be controlled to avoid the noisy neighbor problem. Docker provides configuration options for setting up minimum and maximum limits on CPU, memory and CPU while creating containers. Microsoft provides telemetry solutions at application level and also at the host level. These can capture critical information on the container performance and the host performance. At present these technologies only support Linux-based container hosts.

Dockerfile contents are operated from top to bottom. In order to reduce the build time of an image one should club statements. Docker checks for cached layers for every instruction in the Dockerfile. To increase the re-usability or caching installations not specific to an application, they should be separated out. Ordering of the instructions is also important since any change in the intermediate layer invalidates the remaining cached layers, and Docker starts building...