BudgetPulse.com — Accelerating Time to Market with a DevOps Pipeline in AWS

The challenge

A growing user base, but slow, painful releases.

BudgetPulse.com is an online personal budgeting tool that combines simplicity and functionality. It is sold as a white-labeled solution to leading financial organizations worldwide, with a premium version available to consumers online.

BudgetPulse.com had been deployed on a different cloud provider, and over the previous months, their paid user base had increased substantially. They also needed less downtime around the clock, including weekends. The previous provider had a complicated change management process that made deployments painful: the team performed manual deployments and patching on their own servers, and managing dependencies was a constant challenge.

Due to these limitations and the considerable effort required, deployments could not be performed frequently. New features were delayed in reaching production. It took 5 to 7 days to promote even a simple change from development to production, and every production deployment required 3 to 5 hours of application downtime.

The solution

A re-architected platform with end-to-end automation.

BudgetPulse partnered with Cloudism to replace their previous cloud provider and rebuild their DevOps pipeline to support continuous delivery with zero-downtime deployments. Given our extensive experience with AWS, we recommended AWS Cloud—allowing BudgetPulse to fully leverage cloud computing with the latest technologies, all while satisfying the industry-standard security guidelines for financial applications.

DevOps pipeline architecture for BudgetPulse showing the full software delivery flow: developers push pull requests to GitHub, AWS CodePipeline triggers AWS CodeBuild to build and tag Docker images stored in Amazon ECR — with parallel quality gates running unit tests in Jasmine, static code coverage in SonarQube, and security/license analysis in Nexus IQ. Infrastructure-as-code from CloudFormation and Ansible playbooks deploys to AWS, then AWS CodeDeploy handles database migrations via Sequelize and deploys containers to Amazon ECS. The release candidate is then validated through integration tests, OWASP and Nessus security scans, and load tests before being tagged for production.

Application architecture

BudgetPulse was re-architected with these goals:

Decompose application functionality into a series of container-based microservices that can be maintained, deployed, and scaled independently—significantly increasing application agility and availability.
Leverage an API-first strategy to enable API-based access to other integrating systems, including legacy components.
Review and redesign every application component for maximum reliability and scalability.

Infrastructure

The entire BudgetPulse infrastructure is hosted in AWS, designed and deployed using these best practices:

Provision and decommission infrastructure using AWS CloudFormation.
Provision CloudFormation templates using Ansible.
Use Ansible for patch management when critical updates need immediate attention.
Make the entire infrastructure immutable—rebuilding from the latest AMIs to host the application fresh.
Rebuild environments every two weeks using hardened AMIs to address new security vulnerabilities, with zero downtime.

DevOps pipeline

The DevOps pipeline was redesigned to speed up application development and reduce time between development and deployment:

GitHub — version control where all application code lives.
AWS CodePipeline & CodeBuild — integration server where committed code goes through static analysis, unit testing, build, publish, and deploy stages.
Amazon ECR — artifact manager where Docker images from CodeBuild are published and managed.
AWS ECS via CodeDeploy — Docker images deployed to EC2 instances under the ECS cluster, with high availability and scaling handled automatically.
Checkpoints — places in the release workflow where approvals are required at specific deployment stages.
Quality gates — different levels of testing and security checks that determine whether an artifact is ready to be promoted to the next environment.

The outcome

From multi-day deployments to continuous delivery.

The team achieved greater than 80% confidence in application deployments and the quality of releases, and is now able to perform more than 20 deployments per day across all environments—including production—with zero downtime.

Continuous delivery — any small development or modification gets deployed immediately on commit for testing through automation.
Reduced release timelines — improved developer productivity and innovation.
Quality discipline — code coverage standard improved by ensuring a minimum of 85% coverage for quality delivery.
Infrastructure in minutes — what once took days or a week to provision now takes 25-40 minutes via Infrastructure as Code.
Immutable infrastructure — almost zero configuration management; the entire infrastructure is rebuilt to address vulnerabilities or server failures.
Reduced operations cost — unused servers stopped and restarted as needed; auto-scaling adapts horizontally to traffic.

Accelerating time to market with
a DevOps pipeline in AWS.

Results at a glance