Valve

Hello Retail

Hello-Retail is an award winning serverless, event-driven retail application prototype developed by Nordstrom Technologies.

Being a serverless application, it fits our need for a demo application for evaluating Valve.

We have implemented the variant of Hello-Retail that is implemented by Apernas et al. in the paper: "Secure serverless computing using dynamic information flow control".

They have made some changes to the structure of hello-retail.

As per their repository, there are variants of hello-retail:

1. Structurally modified hello-retail, which we'll call vanilla-hello-retail

2. Hello-retail ported to Trapeze, a system proposed in their paper, which we'll call trapeze-hello-retail

We have modified the vanilla-trapeze-hello-retail, to make it run with Valve. We'll call this variant vanilla-trapeze-hello-retail.

To sum up, there are 3 variants in our work:

1. vanilla-hello-retail

2. trapeze-hello-retail

3. valve-hello-retail

Flows in hello-retail

There are three main flows in Hello Retail, all of which are present in the 3 variants mentioned above:

1. productCatalogApi

   • api

   • builder

2. product-photos

   • 1.assign

   • 2.message

   • 2.record

   • 3.receive

   • 4.success

   • 6.report

3. product-purchase

   • 1.authenticate

   • 2.getPrice

   • 3.authorize-cc

   • 4.publish

Setup and execution:

1. Setup a Kubernetes cluster

2. Deploy the sql service:\

   • Navigate to mysql:

     
     
     
     kubectl apply -f mySqlDeployment.yml
     
     
     
     kubectl apply -f mysql-pv.yml
     
     
     
     

   • Start a mysql client:

     
     kubectl run -it --rm --image=mysql:5.6 mysql-client -- mysql -h mysql -ppassword
     
     

   • Create user and grant priviledge. Add database

     
     create user 'abc'@% IDENTIFIED BY 'xyz';
     
     create user 'abc'@'%' IDENTIFIED BY 'xyz';
     
     create database helloRetail;
     
     

3. Deploy the functions\

• Deploy a function, look below for example:

  cd vanilla-hello-retail/productCatalogApi/builder/
  faas-cli up -f product-catalog-builder.yml 
  

• Call the function, we provide a sample input json in the same folder as the function yml.

  curl -d @sample-input-product.json -H "Content-Type: application/json" -X POST 'http://192.168.99.101:31112/function/product-catalog-builder/product'