Hipster Shop: Cloud-Native Microservices Demo Application

This project contains a 10-tier microservices application. The application is a web-based e-commerce app called “Hipster Shop” where users can browse items, add them to the cart, and purchase them.

Google uses this application to demonstrate Kubernetes, GKE, Istio, Stackdriver, gRPC, OpenCensus and similar cloud-native technologies.

Note to Googlers: Please fill out the form at go/microservices-demo if you are using this application.

Screenshots

Home Page	Checkout Screen

Service Architecture

Hipster Shop is composed of many microservices written in different languages that talk to each other over gRPC.

Find Protocol Buffers Descriptions at the ./pb directory.

Service	Language	Description
frontend	Go	Exposes an HTTP server to serve the website. Does not require signup/login and generates session IDs for all users automatically.
cartservice	C#	Stores the items in the user's shipping cart in Redis and retrieves it.
productcatalogservice	Go	Provides the list of products from a JSON file and ability to search products and get individual products.
currencyservice	Node.js	Converts one money amount to another currency. Uses real values fetched from European Central Bank. It's the highest QPS service.
paymentservice	Node.js	Charges the given credit card info (hypothetically😇) with the given amount and returns a transaction ID.
shippingservice	Go	Gives shipping cost estimates based on the shopping cart. Ships items to the given address (hypothetically😇)
emailservice	Python	Sends users an order confirmation email (hypothetically😇).
checkoutservice	Go	Retrieves user cart, prepares order and orchestrates the payment, shipping and the email notification.
recommendationservice	Python	Recommends other products based on what's given in the cart.
adservice	Java	Provides text ads based on given context words.
loadgenerator	Python/Locust	Continuously sends requests imitating realistic user shopping flows to the frontend.

Features

• Kubernetes/GKE: The app is designed to run on Kubernetes (both locally on "Docker for Desktop", as well as on the cloud with GKE).
• gRPC: Microservices use a high volume of gRPC calls to communicate to each other.
• Istio: Application works on Istio service mesh.
• OpenCensus Tracing: Most services are instrumented using OpenCensus trace interceptors for gRPC/HTTP.
• Stackdriver APM: Many services are instrumented with Profiling, Tracing and Debugging. In addition to these, using Istio enables features like Request/Response Metrics and Context Graph out of the box. When it is running out of Google Cloud, this code path remains inactive.
• Skaffold: Application is deployed to Kubernetes with a single command using Skaffold.
• Synthetic Load Generation: The application demo comes with a background job that creates realistic usage patterns on the website using Locust load generator.

Installation

Note: that the first build can take up to 20-30 minutes. Consequent builds will be faster.

Option 1: Running locally with “Docker for Desktop”

💡 Recommended if you're planning to develop the application.

1. Install tools to run a Kubernetes cluster locally:

   • kubectl (can be installed via gcloud components install kubectl)
   • Docker for Desktop (Mac/Windows): It provides Kubernetes support as noted here.
   • skaffold (ensure version ≥v0.20)

2. Launch “Docker for Desktop”. Go to Preferences:

   • choose “Enable Kubernetes”,
   • set CPUs to at least 3, and Memory to at least 6.0 GiB

3. Run kubectl get nodes to verify you're connected to “Kubernetes on Docker”.

4. Run skaffold run (first time will be slow, it can take ~20-30 minutes). This will build and deploy the application. If you need to rebuild the images automatically as you refactor he code, run skaffold dev command.

5. Run kubectl get pods to verify the Pods are ready and running. The application frontend should be available at http://localhost:80 on your machine.

Option 2: Running on Google Kubernetes Engine (GKE)

💡 Recommended for demos and making it available publicly.

1. Install tools specified in the previous section (Docker, kubectl, skaffold)

2. Create a Google Kubernetes Engine cluster and make sure kubectl is pointing to the cluster.

    gcloud services enable container.googleapis.com
   
    gcloud container clusters create demo --enable-autoupgrade \
        --enable-autoscaling --min-nodes=3 --max-nodes=10 --num-nodes=5 --zone=us-central1-a
   
    kubectl get nodes
   

3. Enable Google Container Registry (GCR) on your GCP project and configure the docker CLI to authenticate to GCR:

   gcloud services enable containerregistry.googleapis.com
   
   gcloud auth configure-docker -q
   

4. In the root of this repository, run skaffold run --default-repo=gcr.io/[PROJECT_ID], where [PROJECT_ID] is your GCP project ID.

   This command:

   • builds the container images
   • pushes them to GCR
   • applies the ./kubernetes-manifests deploying the application to Kubernetes.

   Troubleshooting: If you get "No space left on device" error on Google Cloud Shell, you can build the images on Google Cloud Build: Enable the Cloud Build API, then run skaffold run -p gcb --default-repo=gcr.io/[PROJECT_ID] instead.

5. Find the IP address of your application, then visit the application on your browser to confirm installation.

   kubectl get service frontend-external
   

   Troubleshooting: A Kubernetes bug (will be fixed in 1.12) combined with a Skaffold bug causes load balancer to not to work even after getting an IP address. If you are seeing this, run kubectl get service frontend-external -o=yaml | kubectl apply -f- to trigger load balancer reconfiguration.

Option 3: Using Static Images

💡 Recommended for test-driving the application on an existing cluster.

Prerequisite: a running Kubernetes cluster.

1. Clone this repository.

2. Deploy the application: kubectl apply -f ./release/kubernetes-manifests

3. Run kubectl get pods to see pods are in a healthy and ready state.

4. Find the IP address of your application, then visit the application on your browser to confirm installation.

   kubectl get service frontend-external
   

(Optional) Deploying on a Istio-installed GKE cluster

Note: you followed GKE deployment steps above, run skaffold delete first to delete what's deployed.

1. Create a GKE cluster (described above).

2. Use Istio on GKE add-on to install Istio to your existing GKE cluster.

   gcloud beta container clusters update demo \
       --zone=us-central1-a \
       --update-addons=Istio=ENABLED \
       --istio-config=auth=MTLS_PERMISSIVE
   

   NOTE: If you need to enable MTLS_STRICT mode, you will need to update several manifest files:

   • kubernetes-manifests/frontend.yaml: delete "livenessProbe" and "readinessProbe" fields.
   • kubernetes-manifests/loadgenerator.yaml: delete "initContainers" field.

3. (Optional) Enable Stackdriver Tracing/Logging with Istio Stackdriver Adapter by following this guide.

4. Install the automatic sidecar injection (annotate the default namespace with the label):

   kubectl label namespace default istio-injection=enabled
   

5. Apply the manifests in ./istio-manifests directory.

   kubectl apply -f ./istio-manifests
   

   This is required only once.

6. Deploy the application with skaffold run --default-repo=gcr.io/[PROJECT_ID].

7. Run kubectl get pods to see pods are in a healthy and ready state.

8. Find the IP address of your istio gateway Ingress or Service, and visit the application.

   INGRESS_HOST="$(kubectl -n istio-system get service istio-ingressgateway -o jsonpath='{.status.loadBalancer.ingress[0].ip}')"
   
   echo "$INGRESS_HOST"
   
   curl -v "http://$INGRESS_HOST"
   

Conferences featuring Hipster Shop

• Google Cloud Next'18 London – Keynote showing Stackdriver Incident Response Management
• Google Cloud Next'18 SF
  • Day 1 Keynote showing GKE On-Prem
  • Day 3 – Keynote showing Stackdriver APM (Tracing, Code Search, Profiler, Google Cloud Build)
  • Introduction to Service Management with Istio

---

This is not an official Google project.