All projects
Dstny Logo
Dstny2022–2023

Dstny

A multi-region platform built from scratch, in the cloud and on-prem

Telecom / Cloud / Enterprise

Dstny Logo
Applications Migrated
30+
Global Regions
4
Customer Domains
50+
Enterprise Deployments
10+

Dstny is a telecom and business-communications company, and its infrastructure was fragmented VMs and bare metal. What replaced it is a multi-region Kubernetes platform built from scratch, carrying 30+ applications across production, test and QA in four regions. It had to run in AWS and in on-prem datacentres at the same time.

01

Cloud where it scales, on-prem where calls cannot cross a border

Different customers needed different things. Some required on-prem, some needed their calls kept inside a country's borders, and existing on-prem systems had to keep working alongside the platform. Elsewhere, as in Australia, owning a datacentre was not feasible and the cloud gave fast scaling instead. So the platform runs on OpenShift on-prem and on AWS EKS in the cloud at the same time, matched to each need. Supporting both constrains every choice about ingress, storage and security policy, and it rules out most of the shortcuts a single-runtime platform can take.

Runtimes
EKS + OpenShift
Stack
Terraform, GitOps, GitLab CI
On call
Throughout

02

Fifty customer domains, no manual certificates

Every end customer can bring their own domain, and more than fifty do. Certificates are issued and renewed automatically, and an in-cluster DNS service means onboarding a new domain takes minutes rather than a ticket. Application teams reach their own metrics and logs themselves.

Ingress
ALB/NLB + Traefik
Certificates
cert-manager
Observability
Prometheus, Loki, Grafana

03

The white-label login is a product, and non-engineers operate it

Every enterprise customer gets a login that carries their own brand. I built it end to end, from the backend running inside the authentication service to the front end that loads each customer's look on demand. The shipped code is obfuscated so no customer can be identified from it. Onboarding and account management teams run it themselves across 10+ enterprise deployments, with no engineer in the loop to rebrand a login. I also wrote the training material and ran the Kubernetes sessions with the operations teams who inherited all of this.

Stack
Node.js, React, Keycloak
Deployments
10+

What I owned

Platform Architect & DevOps Engineer

  • Multi-region platform from scratch, replacing VMs and bare metal
  • Platform architecture across four regions, Terraform and GitOps
  • Cloud and on-prem side by side, EKS and OpenShift
  • Self-service observability for application teams
  • Automated certificates for 50+ customer domains
  • White-label login end to end, backend and front end
  • Kubernetes training for the operations teams

Kubernetes / AWS EKS / OpenShift / Terraform / GitOps / Traefik / Prometheus / Grafana / Loki / Keycloak / TypeScript / React / Node.js / Helm / Docker

Next step

Tell me what you're building

Send me a short description of the problem and I'll tell you whether I'm the right fit, what it would cost, and how long it would take. No pitch deck.