Merge branch 'multiModelExample' into 'master'

Added examples for serving multimodel in kf1.3

See merge request !4

serving/multi-model/README.md

+# Multi model serving
+Kubeflow allows you to serve multiple models from a single inference server. You will be able to make use of the same amount of resources to serve multiple models. 
+
+# Create inference service
+The metadata in `inferenceservice.yaml` needs to be edited with a respective which will be used later to access your model. After that 
+
+`kubectl apply -f inferenceservice.yaml`
+
+# Serving Models with the newly created inference service
+You will be able to serve multiple models by replicating `firstmodel.yaml` and just changing the details for your model location.
+Remember change the `inferenceService` name to the one you choose in the previous step.
+
+`kubectl apply -f firstmodel.yaml`
+
+Follow the same steps for `secondmodel.yaml`. You have now deployed 2 models on the same inference service.
+
+
+# Check Status
+To check for the status of these models run `kubectl get trainedmodel` you should now be able to see the 2 newly deployed models with their URLS.
+**REMEMBER** The 'ready' status for both these models should be true inorder for you to access them.
+
+
+# Access the models and inference service
+- AUTH_SESSION is the authentication session cookie obtained in the previous step
+- NAMESPACE is a personal Kubeflow namespace, which can be seen in the top left corner of the UI
+
+- For predicting first model
+```curl -H 'Cookie: authservice_session=AUTH_SESSION' -H 'Host: multi-model-sklearn-example.NAMESPACE.example.com' http://ml.cern.ch/v1/models/multimodel-first-example:predict -d @./input.json```
+
+- For predicting second model
+```curl -H 'Cookie: authservice_session=AUTH_SESSION' -H 'Host: multi-model-sklearn-example.NAMESPACE.example.com' http://ml.cern.ch/v1/models/multimodel-second-example:predict -d @./input.json```
\ No newline at end of file

serving/multi-model/firstmodel.yaml

+apiVersion: "serving.kubeflow.org/v1alpha1"
+kind: "TrainedModel"
+metadata:
+  name: "multimodel-first-example"
+spec:
+  inferenceService: "multi-model-sklearn-example"
+  model:
+    storageUri: "gs://kfserving-samples/models/sklearn/iris"
+    framework: "sklearn"
+    memory: "256Mi"

serving/multi-model/inferenceservice.yaml

+apiVersion: "serving.kubeflow.org/v1beta1"
+kind: "InferenceService"
+metadata:
+  name: "multi-model-sklearn-example"
+spec:
+  predictor:
+    minReplicas: 1  
+    sklearn:
+      name: "sklearn-mm-predict"
+      resources:
+        limits:
+          cpu: 100m
+          memory: 512Mi
+        requests:
+          cpu: 100m
+          memory: 512Mi
+        
\ No newline at end of file

serving/multi-model/input.json

+{
+  "instances": [
+    [6.8,  2.8,  4.8,  1.4],
+    [6.0,  3.4,  4.5,  1.6]
+  ]
+}
\ No newline at end of file

serving/multi-model/secondmodel.yaml

+apiVersion: "serving.kubeflow.org/v1alpha1"
+kind: "TrainedModel"
+metadata:
+  name: "multimodel-second-example"
+spec:
+  inferenceService: "multi-model-sklearn-example"
+  model:
+    storageUri: "gs://kfserving-samples/models/sklearn/iris"
+    framework: "sklearn"
+    memory: "256Mi"