Skip to content

Latest commit

 

History

History
122 lines (90 loc) · 3.95 KB

README.md

File metadata and controls

122 lines (90 loc) · 3.95 KB
Raw

geolocate-scheduler

Test

IoT and other data-intensive systems produce enormous amounts of data to be processed. In this project, we explore the Edge Computing concept as additional computational power to the Cloud nodes, allowing data process computations to occur closer to their source.

Using Kubernetes/KubeEdge to manage the data processing workloads, the geolocate-scheduler aims towards substantial improvements in scalability levels, reducing the request latency and network usage, by scheduling those workloads in Edge Nodes based on the geographical location of the data and resource availability.

Usage

⚠️ First, make sure you have deployed or are running the edge-deployment-controller in the cluster by following the instructions in the project repository. This controller is responsible for the management and reconciliation of the EdgeDeployments, which are the the resource type of our workloads.

Configuration - Environment variables

ALGORITHM defines the method used to select target nodes for workload placement available algorithms are:

  • location(default): where nodes are filtered based on if they have enough available resources, and the target node is selected based on the deployment required/preferred locations,
  • naivelocation: where the target node is selected based on the deployment required/preferred locations
  • random: where any edge node is a good fit for the given workloads

Deployment

kubectl apply -f examples/release_scheduler_crd.yaml

Build custom docker image

docker build -t geolocate-scheduler .

Development

Run the geolocate-scheduler

To develop geolocate-scheduler in a Kubernetes cluster we use ksync to sync files between our local system, and the cluster.

  1. Install ksync. You can follow ksync installation steps here.

  2. Create a deployment where we will run the scheduler by applying the example/dev_scheduler_crd.yaml.

     kubectl apply -f example/dev_scheduler_crd.yaml

  3. If not done before then create a ksync configuration for the current folder.

    ksync create --selector=component=geolocate-scheduler --reload=false --local-read-only=true $(pwd) /code

  4. Start ksync update system

    ksync watch

  5. Run the scheduler in the cluster pod

    kubectl exec -it $(kubectl get pod -n kube-system | grep geolocate-scheduler | awk '{print $1}') -- sh
cd /code
ALGORITHM=naivelocation go run main.go

Lint

go install github.com/golangci/golangci-lint/cmd/[email protected]
golangci-lint ./...

Testing and Coverage

go test --coverprofile=coverage.out ./...
go tool cover -html=coverage.out

Format

go fmt ./...

Manage nodes

The geolocate-scheduler only manages Edge nodes, because the main purpose is to allow application workload to be deployed near the source of data to be processed. Therefore, the node controller filters nodes by 'node.geolocate.io' labels.

  • To add the label

    kubectl label node node0 --overwrite node.geolocate.io=

  • To remove the label

    kubectl label node node0 --overwrite node.geolocate.io-

Deploy workloads

Apply any of the workload examples

  • No location set

    kubectl apply -f examples/workload_no_set_location.yaml

  • Required location

    kubectl apply -f examples/workload_required_location.yaml

  • Preferred location

    kubectl apply -f examples/workload_preferred_location.yaml