Atlas Dev Studio

Ugo-Fresh  is a marketplace specializing in managing the supply chain of fresh products. The company sought external expertise to explore the potential migration of its infrastructure to Kubernetes.

With all internal developers occupied with feature development, we undertook a rapid Proof of Concept (POC) to assess the feasibility of such a migration.

This article details the seamless transition and its impact on scalability and cost-effectiveness.

Proof of Concept (POC) in Two Days

Our team executed a comprehensive POC within just two days. The key components of the POC included the creation of a Google Kubernetes Engine (GKE) autoscale cluster and the integration of Cloud SQL for the database.

This swift implementation allowed Ugo-Fresh.com to test the waters and explore new avenues for scaling their infrastructure efficiently.

Scalability and Cost Savings

The introduction of GKE autoscale clusters empowered Ugo-Fresh.com to scale their infrastructure dynamically, meeting the demands of high-volume periods seamlessly. The Cloud SQL integration provided a robust database solution, enhancing the overall responsiveness of the marketplace. The flexibility to scale up during busy periods and scale down to save costs during low-demand periods marked a significant improvement in operational efficiency.

Local Development Enhancements

As part of the migration recommendation, we advised Ugo-Fresh team to transition from Docker Swarm on localhost to solutions like Minikube or Kind for local development environments. This alignment with the production environment ensures consistency and reduces potential deployment issues.

Additionally, the incorporation of tools like Skaffold or Tilt streamlines local development processes, providing a more efficient and consistent development experience.

Conclusion

The successful POC we created for Ugo-Fresh.com showcased the potential benefits of migrating to Kubernetes. The rapid implementation of a GKE autoscale cluster and the utilization of Cloud SQL not only enhanced scalability but also provided cost-saving opportunities.

The recommendation to optimize local development environments with solutions like Minikube or Kind, along with tools like Skaffold or Tilt, further demonstrates our commitment to delivering comprehensive solutions that align with clients’ evolving needs.

As Ugo-Fresh.com continues to grow, the groundwork laid by this POC positions the project well for a scalable and efficient future.

GetWebStack is a CLI designed to streamline project setup. As a start-up, it was looking for a robust infrastructure solution. The project focus is to drastically reduce initial setup times from weeks to less than an hour.

GetWebStack.com required an efficient, scalable, and cost-effective architecture.

In this case study we dive into the intricacies of the implemented system that seamlessly integrates frontends, APIs, databases, and a diverse set of tools.

Architecture Overview

The project GetWebStack.com boasts a sophisticated architecture comprising:

• two frontends
• an API
• a CLI
• a database
• a repository for pre-built micro-apps

The static frontend, implemented in Angular and deployed on Google Cloud Storage, handles landing pages and unauthenticated user content. The second frontend, also Angular-based, caters to user project management. Both frontends are pre-rendered, contributing to optimal load time.

The API, written in NodeJs, leverages the serverless capabilities of Google Cloud Run, ensuring scalability and efficiency. Google Cloud SQL serves as the primary database, while external Redis facilitates session management.

Key Components and Implementations

• CLI Deployment: The NodeJs-based CLI, a cornerstone of GetWebStack.com, is deployed to NPM, providing users with a seamless experience in accessing pre-built micro-apps.
• Micro-Apps Repository: A repository for pre-built micro-apps, stored on Google Cloud Storage, fuels the quick setup provided by GetWebStack.com.
• Build Pipelines: CircleCI orchestrates the build pipelines, managing the Angular and NodeJs templates. This includes repository setup on GitHub, installation of templates, and implementation of CircleCI pipelines and hostnames.

Infrastructure Setup

The comprehensive setup, completed within two weeks, involved steps, such as creating repositories on GitHub, template installations for Angular and NodeJs, and the configuration of CircleCI pipelines and hostnames.

Cost-Efficiency

Given the current low user volume, the monthly infrastructure cost for GetWebStack.com is impressively under 20 EUR. Leveraging serverless solutions like Google Cloud Run, storage solutions, and efficient deployment pipelines contributes to this exceptional cost-effectiveness.

Conclusion

The work we did for GetWebStack.com has resulted in a robust, scalable, and cost-efficient infrastructure that aligns with the initial CLI’s vision of simplifying project setups.

As GetWebStack.com continues to grow, the implemented architecture ensures adaptability and scalability, showcasing our commitment to delivering optimal solutions for evolving client needs in the dynamic landscape of modern development.

We pride ourselves on our ability to optimize infrastructure and operations for our clients.

In this case study, we present the improvements we made for Booktr.ee resulting in a sixfold decrease in monthly costs.

Booktr.ee is a marketplace for buying and selling used books between private individuals. Leveraging our expertise in DevOps, we completely transformed Booktr.ee’s infrastructure. This not only reduced costs but also significantly improved the overall application performance.

Challenge

Initially, Booktr.ee operated on a Kubernetes cluster on Google Cloud (GKE). However, the startup’s fluctuating traffic patterns, particularly during periods of no usage, led to high infrastructure costs, reaching 300 EUR/month.

The challenge was to find a cost-effective solution without compromising performance.

Solution

After careful analysis, we recommended a transition to a serverless architecture, specifically opting for Google Cloud Run. This move allowed Booktr.ee to pay only for the resources consumed during active periods, eliminating costs during low or 0 traffic periods. Additionally, the free tier associated with Google Cloud Run offered further cost savings.

Implementation

To expedite the migration process, we utilized Docker images already in use on the Kubernetes cluster. The separation of databases with Cloud SQL and an external Redis resource ensured a seamless transition. Within just one week, we successfully implemented the serverless solution.

While the serverless approach proved highly cost-effective, it presented challenges. Those consist particularly of cold starts and the delayed startup time for frontend and backend services.

To address this, we focused on optimizing startup times in the API by separating the update procedure from the main API. For the Angular-based frontend, we implemented server-side rendering, ensuring a swift initial load.

Results

The result was remarkable. We achieved an immediate decrease in infrastructure costs from 300 EUR/month to just 50 EUR/month. The implementation of serverless architecture not only slashed Booktr.ee’s monthly infrastructure costs but also improved overall system efficiency.

Eager to try this solution? Check out our article on the  requirements of serverless technologies for more insights.

Conclusion

We successfully transformed Booktr.ee’s infrastructure, showcasing our commitment to delivering cost-effective and high-performance DevOps solutions.

Our strategic shift to serverless architecture on Google Cloud Run, coupled with optimizations for startup times, resulted in tangible benefits for Booktr.ee.

As we continue to drive innovation in DevOps, our focus remains on providing tailored solutions that empower businesses to thrive in the ever-evolving digital landscape.

Driven by our passion for startups, we collaborated with  Marmelado.io from the idea stage to the beta version. Marmelade.io is a goods exchange marketplace connecting an unlimited number of people. It operates by identifying exchange circles. The beta version includes three categories: clothes, home appliances, and tech devices.

The goal was to provide a working MVP with high availability and scalability, aligning with our ongoing mission to deliver efficient and cost-effective DevOps solutions.

While we also managed the platform’s development, this article will focus on how we created the infrastructure at minimal cost.

Challenge

Marmelade.io aimed to launch an MVP encompassing:

• 3 landing pages for each product category (clothes, home appliances, and tech devices)
• one main landing page
• an authenticated user section
• a waiting list for pre-launch account creation.

The main challenge was to do this with low infrastructure costs. We also had to ensure a smooth user experience.

Solution

To address Marmelade.io’s requirements, we devised a strategic solution leveraging serverless architecture. Read this blog post to find out more about the requirements of serverless technologies.

The chosen tech stack included a frontend written in Angular with server-side rendering, prerendering static content wherever possible, and an API built using ExpressJs.

Given the expected low initial user volume, we opted for Google Cloud Run for the API and Firebase for the frontend.

Implementation

Our implementation phase spanned two weeks. During this time, we set up repositories for the Angular frontend and Express server. We also created a deployment pipeline using CircleCI and configured the project on Google Cloud.

Finally, we went for a Cloud SQL instance to serve as the database. This streamlined setup allowed for rapid development and deployment.

Infrastructure Cost

The standout feature of our approach was the nearly negligible infrastructure cost. The serverless architecture, coupled with the free tiers offered by Google Cloud Run and Firebase, ensured that Marmelade.io’s costs remained close to zero.

The only associated monthly fee came from the Cloud SQL instance. This one proved to be a minimal investment for the overall infrastructure.

Results

The swift and cost-effective implementation of the MVP enabled Marmelade.io to launch with confidence. The dynamic combination of Angular with server-side rendering, ExpressJs, Google Cloud Run, Firebase, and CircleCI provided a robust and scalable foundation.

The waiting list feature allowed users to express interest pre-launch, building anticipation for the marketplace.

Conclusion

We highly enjoyed working on this project. It proved our commitment to providing tailored solutions that align with clients’ specific needs.

By strategically combining serverless technologies, cloud services, and an efficient deployment pipeline, we enabled Marmelade.io to realize its MVP vision with minimal infrastructure costs. On top of this, we created a good infrastructure base for future growth.

As Marmelade.io continues to grow, we remain poised to support their evolving DevOps requirements. We continue helping the project scale seamlessly in the dynamic digital landscape.