Optimizing a b2b software platform infrastructure

Summary

We revamped a client’s infrastructure into a robust, fully automated system with Kubernetes, ensuring high availability, disaster resilience, and secure operations. The solution improved deployment processes, enhanced monitoring, and kept the cost low compared to managed Kubernetes solutions.

blocks

Services used

hetzner
kubernetes
elastic search
logstash
kibana
prometheus
ceph
rook
rancher
vault
sftpgo

Areas of work

Deployment

Before: The deployment process was complex and manual, with uncertainty about pipeline correctness.

After: Streamlined deployment using Kubernetes, Gitlab CI, Terraform, ArgoCD, and Helm Charts for visibility and convenience.

High Availability & Disaster Recovery

Before: The infrastructure was not resilient to data center failures, and disaster recovery was a major concern.

After: Implemented Kubernetes with geographically distributed nodes, Rook Ceph for high availability, and ensured applications run even if a data center fails.

deployment
network

Network Configuration

Before: The architecture was tied to a single, unchangeable IP address for external communication.

After: Configured the network subsystem with Cilium to maintain a specific IP for external services, allowing future customization

deployment

Backups

Before: Daily backups with a 7-day retention period were insufficient.

After: Developed custom backup solutions for persistent volumes and Postgres databases, with backups sent to S3 storage in encrypted form and a 3-month retention period.

Documentation

Before: No documentation existed, making it difficult to manage a black-box infrastructure.

After: Comprehensive documentation was created for the infrastructure and code delivery process.

Monitoring & Logging

Before: No centralized monitoring or log collection.

After: Set up ELK-based log collection and Prometheus/Grafana for centralized monitoring.

servers

Other

Security. We kept encrypted connections secure and automated while transferring all security systems to the new infrastructure. Cert Manager was implemented for automatic certificate issuance. We configured Traefik to eliminate unencrypted access and used Hashicorp Vault for secure secret storage. We also increased the security level by implementing a web application firewall.

Cost Efficiency. Managed Kubernetes solutions are very pricy, so keeping the expenses for the new infrastructure at the lowest possible level was a good bonus.

Results

99,95%

infrastructure

reliability

150 EUR

resource

cost

25%

lower maintenance fee

compared to competitors

Client’s Outcome

reliable systems

Reliable System:

Achieved a reliable, fully automated, crash-resistant system.

blocks

Scalability:

Ensured scalability and secure service releases.

monitoring

Effective Monitoring:

Enabled effective monitoring and logging systems.

A list of technical details

expand arrow

Related posts

Well, shall we get started?