Mailstep: Clear, Modern, and Robust AWS Infrastructure for Logistics
When you handle logistics for hundreds of e-shops, you need a reliable, modern infrastructure that also provides the data you need for invoicing. That's exactly what we delivered for Mailstep.
The Czech company Mailstep processes more than 30,000 e-shop parcels every day. Through its own application, it manages integrations with various e-commerce platforms and carriers, and, most importantly, complete order fulfillment for online stores. In 2023, the company hit the limits of its infrastructure, which was unnecessarily complex, fragmented, and outdated. Mailstep turned to us, and within two weeks we had taken over DevOps and started working on optimization.
„External collaborations are often a gamble, especially in an area as sensitive as DevOps, where the partner essentially gets the 'keys to the castle'. With Cookielab, we've always felt that our security and privacy come first. They helped us set up processes for managing developer and technical role access to AWS resources, and they truly took ownership of our infrastructure — so much so that it felt like they were part of our internal team.“
The story
When we took over DevOps, our first step was to analyze the state of the infrastructure. We discovered that Mailstep had around 26 different AWS accounts even though everything served the same purpose — managing fulfillment. On top of that, it was running on relatively old versions of various systems. We decided to consolidate the infrastructure to avoid technical issues and reduce overall DevOps and development costs.
Our first move was to consolidate AWS accounts from more than twenty down to four. Two handle technical matters, one provides an overview of the infrastructure's status, and the fourth manages billing, domain administration, and more. We completed the final adjustments in October 2024, roughly 18 months after taking over the infrastructure.
We migrated data without disrupting Mailstep's services for its clients. We also prepared a unified dashboard, giving everyone a clear view of the infrastructure status, third-party service integrations, and more. We then set up monitoring and alerting so Mailstep would be notified of any issues before their customers noticed.
The challenges
The biggest challenge was the sheer complexity and fragmentation -26 different AWS accounts, outdated system versions, and no unified view of the infrastructure. Everything had to be consolidated and modernized without disrupting service for Mailstep's hundreds of e-shop clients. We completed the migration to 4 accounts while maintaining zero downtime.
What's next
By taking over infrastructure management, we simplified and reduced the cost of working with AWS. Now in 2025, we're completing further optimizations and reviewing each service to check how much server capacity it actually uses compared to how much it should. The goal is to continue reducing costs while introducing modern, security-focused approaches, including an automatic scaling system that adapts to current needs.
The Czech company Mailstep processes more than 30,000 e-shop parcels every day. Through its own application, it handles integrations with various e-commerce platforms and carriers, and complete order picking and fulfillment for online stores. In 2023, the company hit the limits of its infrastructure, which was unnecessarily complex, fragmented, and outdated. Mailstep turned to us, and within two weeks we had taken over DevOps and started working on optimization.
What we delivered
„Mailstep came to us with a big challenge — the infrastructure was fragmented, complex, and full of tech debt. For us, it was a great opportunity to show how quickly we can simplify, clarify, and modernize things. In just a few months, we built a robust, efficient, and secure solution that was ready for further growth. For our team, it was actually a piece of cake, because this is exactly the kind of work we enjoy.“
Technologies
The story
When we took over DevOps, we found that Mailstep had more than 26 different AWS accounts, all serving one goal — managing fulfillment. The original setup used outdated and unmaintained infrastructure management tools, for example FluxCD version 0.32 vs the then-current 2.0.
First, we consolidated the accounts and replaced the old, complicated deployment system with a simpler and clearer approach. We went from more than twenty AWS accounts down to four: non-production, production, tooling (observability / CI pipelines), and a management account for billing and domain administration.
We moved away from Flux and chose the standard Helm release approach, temporarily deprioritizing GitOps to focus resources on bigger problems. We migrated data without threatening service availability and prepared a single place where the client can monitor the entire stack.
We built infrastructure monitoring and metrics on Grafana, Mimir, and Tempo OSS — the most cost-effective solution for the customer's needs. To improve full-text search, we replaced Grafana Loki with OpenSearch. For alerting, we chose Prometheus Alert Manager.
What's next
Now in 2025, we're finalizing optimizations and gradually reviewing each service. We're delivering a ready-to-use system for automatic workload scaling built on KEDA and Karpenter. The long-term plan is to prepare an automated platform where developers have the tools they need for everyday work — improving Developer Experience, speeding up development, and reducing dependency on other teams.