Seldon Deployment and Canary updates

Launch a simple seldon deployment and update model by canarying

Iris Model

iris

We will show:

  • Deploying a pretrained sklearn iris model
  • Loadtesting the model
  • Looking at request logging
  • Canarying a new XGBoost model
  • Load test canary model
  • Promote the canary model

Deploy Model

deploy1

Use the model url:

gs://seldon-models/sklearn/iris

deploy2

Wait for the deployment to be running:

deploy3

Enter the deployment details screen

detail

Start Load Test

Complete the load test wizard:

Use the request.json file in this folder:

{
    "data": {
	"names": ["Sepal length","Sepal width","Petal length", "Petal Width"],
	"ndarray": [
	    [6.8,  2.8,  4.8,  1.4],
	    [6.0,  3.4,  4.5,  1.6]
	]
    }
}

load1

When running you should see metrics on dashboard:

load2

View Request Logs

Enter the request logs screen to view requests.

logs

Create Canary

Create an XGBoost canary model using the saved model at:

gs://seldon-models/xgboost/iris

canary

Rerun the load test and you should see metrics for both default and canary models.

canary-load

Promote the XGBoost Canary to be the main model.

promote

Delete Model

Delete the model

delete


Last modified October 29, 2019: Reorganizing project structure and adding WIP image explanations (8bfc877)