Title: Universality enhancement of Unikernel

 

Technical Area: System Software

 

Background

In order to solve the problem of virtual machines resources’ waste and the safety problem of containers, the industry has been increasingly using Unikernel technology as the solution in the past two years. In addition, its small, simple, safe, and efficient features make it very suitable for the current popular micro service architecture.

 

The biggest problem at the moment is that the ecology is not yet perfect, and none of the Unikernel claims to have been put into the actual production environment. The biggest problem that hinders its ecological progress and industrial landing is that the compatibility and universality to current application, especially in the container cloud environment. Currently, most containers are still complex and have multiple processes using different language. If they can be released still as OCI image without compilation, especially for compilation type language, and still scheduled by docker/k8s, and the container instances can still share the same parts of the OCI images for both memory and disk resources, most of them will be easier and more willing to run or migrate to Unikernel platform. Unfortunate, there isn’t any Unikernel open source project satisfies all of these targets, even two above targets. At last, there isn’t any Unikernel project behavior are proved to be totally same as Linux kernel, because nobody has already successfully migrated current existing application to the production environment.

 

Target

We hope one Unikernel implementation can be researched focus on the compatibility and universality without the feature of good safety, performance and fast booting. The new Unikernel implementation should include but not limited to below universality factors:

 

Related Research Topics