Skip to content

Enables a Kubernetes federation to automatically manage multi-cluster infrastructure

License

Notifications You must be signed in to change notification settings

arjav-desai/navarkos

 
 

Repository files navigation

wercker status Go Report Card

Navarkos

Navarkos acts as a mediator between demand (deployment replicas) and supply (federated clusters). When you are deploying an application to a group of clusters, you need a way to distribute replicas of that application to multiple clusters. Navarkos helps in re-balancing replica allocation among clusters to optimize usage or load by analyzing:

  • what clusters are there and how they are currently used for applications,
  • how much capacity must be allocated for each cluster, and
  • which cluster is best suited for the application.

Navarkos is best used in tandem with cluster-manager to help re-balance demand by performing life cycle operations on clusters, which is achieved either by provisioning more capacity (Provision/Scale up) or removing excessive allocated capacity (Scale Down/Shutdown) i.e. its provides application-aware infrastructure.

Before You Begin

The default Scheduling provided as part of the Federation project distributes replica equally among Federated clusters, but it doesn't provide application-aware infrastructure.

Navarkos works on re-balancing, after initial equal distribution and persists that by updating ReplicaAllocationPreferences annotation on the deployment.


Navarkos with cluster-manager

Figure:1 Navarkos with cluster-manager Architecture

The above architecture diagram shows how Navarkos works with cluster-manager. Navarkos communicates with cluster-manager for life cycle operations by requesting state change in the form of a cluster annotation.

When used with cluster-manager, Navarkos provides following advantages over default scheduling:

  • Consistency: Cluster capabilities consistency (ingress, nodes, network)
  • Manageability: Automated cluster life-cycle management
  • Portability: Trivially move workloads from one set of clusters to another
  • Cost: offline when idle, scale workload with application awareness
  • Compliance: Ensure application is run where and how it needs to be
  • Multi-cloud: Manage application independently of cloud provider
  • Global Scale: Run your application wherever there is demand
  • Hybrid: Allows for private & public clusters to work together dynamically

Build

requirements:

  • Go 1.8.x (Note: Go 1.9 with glide will have nested dependency errors on glide up -v)
  • Glide 0.12+ (0.12.3 and 0.13.0 are the tested versions)
  • Docker (Multi-stage builds requires Docker 17.05 or higher on the daemon and client.)

get source:

  • export K8S_INCUBATOR_DIR="$GOPATH/src/github.com/kubernetes-incubator"
  • mkdir -p $K8S_INCUBATOR_DIR
  • cd $K8S_INCUBATOR_DIR
  • git clone https://github.com/oracle/navarkos

build:

  • make build-local will clean up, builds the navarkos binary and runs locally.
  • make build-image will invoke 'build-local' and builds navarkos docker image, which is suitable for deploying to Kubernetes.
  • make build-docker will cleanup and builds docker image containing navarkos binary, which is suitable for deploying to Kubernetes. Resultant docker image is produced using the multi-stage build docker file.
  • make check will fmt, vet and unit test the code.

Deploy

Follow deploy instructions to deploy Navarkos.

How to use

Learn how to use Navarkos with an example.

Community, discussion, contribution, and support

Learn how to engage with the Kubernetes community on the community page.

You can reach the maintainers of this project at:

  • Slack: tbd
  • Mailing List: tbd
  • OWNERS file has the project leads listed.

Contributions to this code are welcome!

Refer to CONTRIBUTING.md and LICENSE on how to contribute to the project. Refer to RELEASE.md about the process of release of this project.

Kubernetes Incubator

This is a pending proposal for Kubernetes Incubator project. The project was established 2017-12-05. The incubator team for the project is:

  • Sponsor: tbd
  • Champion: tbd
  • SIG: tbd

Code of conduct

Participation in the Kubernetes community is governed by the Kubernetes Code of Conduct.

About

Enables a Kubernetes federation to automatically manage multi-cluster infrastructure

Resources

License

Code of conduct

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Go 98.2%
  • Makefile 1.2%
  • Other 0.6%