choyiny / reliable_blog

A simple blog stack to demonstrate Google's Site Reliability Engineering principles.

Deploying Reliable Software

A CSCD94 - Computer Science Project with Professor Thierry Sans.

Introduction

An extension to the existing “Programming on the Web” course at the University of Toronto Scarborough. This research project aims to follow the two books by Google, Site Reliability Engineering and Site Reliability Workbook. There are two main objectives:

• What is the Google way of deploying software used by millions of people?
• How to maintain high availability while allowing for frequent software updates?

Services

• Main Website: https://deployingreliable.software
• Backend Service: https://backend.deployingreliable.software
• Kibana Dashboard: https://kibana.deployingreliable.software
• Elasticsearch: https://elastic.deployingreliable.software

Development Setup

Prerequisites: Docker and Docker Compose installed in your system.

1. To start all servers in development mode, simply run

$ docker-compose up -d

Production

• We use ansible scripts to push applications to our production servers hosted on Google Cloud. See ansible/README.md.

Technologies

We try to use a range of technologies in this demo. Here are the list of technologies used at some time in this project.

Deployment

"we want systems that are automatic, not just automated. In practice, scale and new features keep SREs on their toes."[1]

• Github Packages: Docker image repository
• Github Actions: Continuous integration and Deployment
• Ansible: Playbooks to automate deployment
• Docker Swarm: Docker across many hosts
• Kubernetes: Container orchestration

Application Stack

"To provide a model of how a service would hypothetically be deployed in the Google production environment"[2]

• Ruby on Rails: Backend server REST framework
• Angular 8: Frontend framework
• PostgreSQL: Relational Database
• nginx: Reverse proxying for applications (Kubernetes Ingress)

Packaging applications

• Docker: Deploy applications cross-platform in a standardized environment

Monitoring Stack

"The four golden signals of monitoring are latency, traffic, errors, and saturation. If you can only measure four metrics of your user-facing system, focus on these four."[3]

Prometheus-based

• Grafana: Dashboard to show statistics and alert if thresholds are exceeded.
• Prometheus: A time series database to store data.
• node_exporter: A companion for Prometheus to export server statistics.

Elasticsearch-based

• Kibana: Dashboard to visualize data from Elasticsearch
• Elasticsearch: Full text search engine to store log data
• Elastic APM: Application performance monitoring server (*uses elastic-apm agent in Rails application)

Others

• Sentry: An external service to track application errors.