Docker containers are becoming more and more popular as a deployment platform for enterprise applications. However, the impact of these applications running in Docker containers on virtualized infrastructures has been studied poorly. There is a new white paper available that uses the Weathervane open source performance benchmark to examine the performance of an enterprise web application running in Docker containers on VMware vSphere 6.5 (VM) virtual machines. The results show that an enterprise web application can run in Docker in VMware vSphere not only without performance degradation, but with even better performance than installing Docker on bare-metal.
Weathervane is used to assess the performance of virtualized and cloud infrastructures by deploying the enterprise web application in the infrastructure and then load the application. The tests discussed in the document use three deployment configurations for the Weathervane application.
- VM without Docker containers: The application runs directly on guest operating systems in vSphere 6.5 virtual machines without Docker containers.
- VM with Docker containers: the application runs in Docker containers that run on the guest operating systems in vSphere 6.5 virtual machines.
- Bare-metal with Docker containers: the application runs in Docker containers, but the containers run on an operating system that is installed on a standard physical server.
The figure below shows the maximum results achieved when running the Weathervane test in three configurations. The results using Docker containers are impacted by the configuration settings, which are discussed in detail in this document.
Some important points that should be noted in these results:
The performance of an application using Docker containers in vSphere 6.5 virtual machines is almost identical to that of the same application running on virtual machines without Docker.
An application running in Docker containers in virtual machines outperforms the performance of the same application running in Docker containers on a bare-metal server by about 5%. Much of the performance gain can be attributed to the complex algorithms used by the vSphere 6.5 scheduler.
The results reviewed in the document, as well as previous Docker performance studies on vSphere, show that vSphere 6.5 is the ideal platform for deploying applications in Docker containers.
You might like: The Kubernets Test on Bare-Metal
