Skip to content

Latest commit

 

History

History
101 lines (64 loc) · 5.72 KB

deployment-instructions.md

File metadata and controls

101 lines (64 loc) · 5.72 KB
Raw

Product Recommendations Preconfigured Solution Deployment Instructions

This document outlines step-by-step what you need to do to deploy the recommendations preconfigured solution.

If you don't already have an Azure account, you will need to create one as this solution deploys a fully functioning recommendations-service to your subscription.

Installing the preconfigured solution

  1. Go to the Product Recommendations Template on the Cortana Analytics Gallery.

  2. Enter the Deployment Name, select the Subscription where you would like to install the solution, the Location (Region) for the deployment. You may also enter an optional Description for your deployment.

  3. Click Create

Deployment Step 1

Provide configuration parameters.

You will select a few parameters that will impact the size of the resources that will be created.

  1. The solution will create an Azure Storage account where the models will be stored. The storage account is also used to store model-metadata and any state required for the solution to work. Specify the type of replication that you would like on your storage account.

  2. The solution will run as an Azure WebApp. You will need to select the application hosting plan. Note that you can always change the plan from Azure Portal even after you've deployed the service.

Important: The selected plan will determine the size of your machine, and therefore the number of models you can concurrently train and the latency of get-recommendation requests. You can use these benchmarks to help you choose the right one to start with and later on adjust the plan according to your specific needs and observed latencies

If you want to increase the size of the scale up or scale out after deployment, you can do that from the Azure Portal.

Deployment Step 2

  1. Click Next

Using your newly created solution

Once all the resources have been deployed, you will be provided two keys (adminKey and recommendKey) that can be used to access a newly create RESTful endpoint that you can use to train models, and get product recommendations from those models. Take note of those keys, as you will need them later on.

The adminKey is the key that can be used for all API operations and gives full error stack on any internal errors, the recommendKey can only be used to get recommendations, so this is the key you would use on the client or website requesting recommendations.

Those keys can also be found in the Azure Portal, as Application Settings for the newly create AppService.

Congratulations! You now have a recommendations service you can use to train models. Take a look at the Getting Started Guide to learn how to create your first model. If you want to learn abut the APIs exposed you can also take a look at the API Reference.

Post Deployments

To deploy new version of the code on an already deployed solution, there are several possible ways. Two of the many approaches are described below:

1. Ad-hoc deployment using Visual Studio

Manual deployments can be performed from visual studio. More details here.

Pre-requisites
  1. Visual Studio 15 or greater
  2. Local clone / copy of the source code solution - https://github.com/Microsoft/Product-Recommendations/tree/master/source
  3. Azure Subscription Access on which solution was originally provisioned.
Steps
  1. Download the publishing profile from the deployed App Service via Azure portal Download Publish Profile
  2. Open the Recommendations solution - Recommendations.Core.sln
  3. Right click on the Recommendations.WebApp project from the Solution Explorer and click Publish
  4. Import the publishing profile from step 1. Refer to Creating a Publish Profile.
  5. Publish the changes. Refer to Previewing Changes and Publishing the Project.

2. Continous deployment from Azure Portal

An automated pipeline can be setup to code-build-deploy new changes. More details are here.

Pre-requisites
  1. Azure Subscription Access on which solution was originally provisioned.
Steps

  1. Fork the git hub project

  2. Open Deployment options in the deployed App Service on Azure Portal. Deployment options

  3. Choose source as Github and select the project Production Recommendations (forked) and hit OK.

  4. Important - Since the github project contains multiple solutions, we need to set the one we want to use explicitly. This can be done by adding a new setting in Application Settings of the App.

    Key - PROJECT

    Value - source/Recommendations.WebApp/Recommendations.WebApp.csproj

    Application Settings

  5. Any changes pushed to the forked branch will be automatically built and deployed. Post-Setup