Skip to content
/ research Public

A place where I host some research content I've produced

Notifications You must be signed in to change notification settings

luxas/research

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 

Repository files navigation

research

A place where I host some research content I've produced.

Content

Encoding human-like operational knowledge using declarative Kubernetes operator patterns

In recent years, the cloud native paradigm has emerged as an effective way of managing server infrastructure, with over 100,000 open source contributors and users such as Apple, CERN, Deutsche Telekom, JD.COM, Nokia and Spotify. Cloud native is a mindset; a set of patterns and practices that provide for efficient, scalable, and resilient server infrastructure control.

Although cloud native has a definition and has been implemented widely, accessible context on why cloud native converged to its present state is scarce. Consequently, if the reasons for a pattern is not known, the pattern might not be implemented correctly, or, worse, at all. The aim of this thesis is to gather this context and showcase the benefits of cloud native implementation in Kubernetes operators. In other words, the goal is to identify methodologies for being able to scale systems, even beyond human cognitive limits, while still keeping the system in a reliable state.

Through review of articles relating to server infrastructure control and open source cloud native resources, this thesis finds that some control problems are noticed mostly at large scale, which can lead small-scale users into a false sense of control. The problems include the impact of inevitable entropy increase, randomness, failures and limited information transfer rate. Through codifying human-like operational knowledge, Kubernetes operators can effectively combat these problems and allow for greater resource, knowledge, and monitoring scalability.

Kubernetes operators form a novel programming model, allowing a shift from humans managing servers to servers managing servers, analogously to the Industrial Revolution. This empowers users to once codify the control mechanism, deploy it using a declarative abstraction layer, then finally, benefit from both system scalability and reliability.

File-specific Licensing

No general, repo-wide license for now. Here's what that means. Instead, each file or directory is explicitly licensed (e.g. using some kind of Creative Commons license).

No Contributions (yet!)

In the future, I'd like to make this repo collaborative, such that I can accept contributions and check in the LaTeX source code. Sadly, I cannot at the moment, with this setup. But that will come at a later stage.

About

A place where I host some research content I've produced

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published