Skip to content

Latest commit

 

History

History
83 lines (70 loc) · 8.41 KB

GETSTARTED.md

File metadata and controls

83 lines (70 loc) · 8.41 KB
Raw

Get Started with the Baseball Machine Learning Workbench

There are multiple ways to get started with the Baseball Machine Learning Workbench, for non-technical users to professional software developers.

Type Azure Required Description
Live Demo Site No Get started exploring the workbench directly on the web.
Run the Docker container locally in your environment No Run the workbench container locally in your Docker environment. Allows for offline exploration using ML.NET machine learning models inside of a Docker container.
Publish Docker Container to the Azure Cloud Yes Publish the workbench Docker conatainer to Azure using Azure Container Instances (ACI)
Publish to Azure App Instance with SignalR Yes Publish the workbench application using the Visual Studio IDE with more advanced options. Use SignalR to scale the communication between the Blazor client and the server.

1. Live Demo Site

The easiest way without performing any technical tasks is to simply use the live demo application located here:
Live Demo Web Site Address: https://aka.ms/BaseballMLWorkbench
Live Demo Web Site (Azure Container Instance): http://baseballmachinelearningworkbench.eastus2.azurecontainer.io
Instructions (show me) guide is provided inside the workbench site.

2. Run the Docker Container locally in your own environment

The complete Baseball Machine Learning Workbench is containerized using Docker. This allows you to run the entire web application locally by using Docker. Linux containers are used. To get started:

  • Ensure you have Docker Engine installed on your Windows, macOS or Linux system. More information located here: https://docs.docker.com/engine/install/
  • Using a command terminal, pull down the Baseball Machine Learning Workbench Docker Container using this command (this will download the docker container locally):
    docker pull bartczernicki/baseballmachinelearningworkbench
  • Verify the Docker image has been pulled down locally, by running this command:
    docker images bartczernicki/baseballmachinelearningworkbench
  • Using a command terminal, run the the Baseball Machine Learning Workbench Docker Container using this command:
    docker run -it --rm -p 8080:80 --name baseballmachinelearningworkbench bartczernicki/baseballmachinelearningworkbench:latest
  • Once you have verified your Docker container is running, using your browser navigate to: http://localhost:8080

3. Publish Docker Container to the Azure Cloud using Azure Container Instances

Since the Baseball Machine Learning Workbench is containerized and published on a public DockerHub repository, you can easily use run the container on Azure. Azuer Container Instances (ACI) allow you to run Docker containers in your own Azure environment. You can do sophisticated deployments using SSL, Application Gateways, VNETs etc. Below is a simple way to get started to publish the Baseball Machine Learning Workbench using ACI

  • Navigate to Azure Portal: https://portal.azure.com
  • Select Create a Resource from the top navigation menu
  • Type in Azure Container Instance inside Azure Marketplace
  • Select the Container Instances Marketplace offer; and select Create on the next screen
  • Fill out the Resource group information. You can name the Container name anything you like; select a Region
  • Select the Image Source: DockerHub or other registry option. The Image name is bartczernicki/baseballmachinelearningworkbench:latest. The OS Type is Linux
  • You can change the size of the image if you like. Below is a screenshot of how this should look like filled out.

  • Select Next: Networking to fill out the Network options.
  • Provide a DNS name label name. This will allow you to access the workbench via a FQDN rather than an IP address. Note: This is a public DNS name scoped to the deployment region, so your FQDN will have to be unique. The screenshot below should show an example of how this will look like.

  • There are more advanced options that could be set, however for now the deployment is ready. Select Review + Create. Your deployment will be validated, if it has been filled out correctly click Create. It will take a few seconds to deploy the Baseball Machine Learning Workbench container to Azure.
  • After your deployment succeeds, you can select Go to resource, where you will have the IP address and the FQDN (DNS) name where you can access the web application. Using your browser, access the application using either the IP address of FQDN. The screenshot below shows where you can find the IP & FQDN in your ACI resource deployment:

4. Publish to Azure App Instance with SignalR

This more advanced publishing option allows you to build the workbench source code, target your desired architecture & frameworks and add more advanced dependencies such as App Insights Instrumentation and scaling out messaging with SignalR. In order to continue, you will need the Visual Studio 2019 IDE and local Git installed.

  • There are several publish targets available. Select the Azure App Service publish target. Select Create New if you don't have an existing instance. Select the Create Profile button.

  • Fill out the App Service details: Subscription, Resource Group, Hosting Plan, Application Insights (optional). Finally select the Create button.

  • You will notice there is a yellow marker next to our Publish button providing a warning, that there are dependencies missing that could improve the performance application. Scroll down and select the + Add button, this will bring up the Add dependency dialog. Select the Azure SignalR Service option and select Next. In the next window you can create a new Azure SignalR Service instance or select an existing one to route messages from the workbench. Note: Azure SignalR has a FREE tier that can be used.

  • Note the variety of deployment options (target frameworks, Deployment modes that can be changed). Click the Publish button, this will start the process of building the solution and pushing it out to Azure. Once publishing has completed, you will see a successful message in the Output window. It will also provide you with a web link for you to immediately test the workbench to ensure it is has been successfully deployed.