Azure infrastructure & DevOps for Rebll Tickets
Moving Rebll Tickets onto codified, self-healing Azure infrastructure — with migration-safe, health-verified releases and payment-grade observability.
When you take real money at scale, the infrastructure is as important as the app. We made Rebll's reproducible, self-verifying and observable — so releases are boring and problems are loud.
We took Rebll Tickets — a revenue-critical events-ticketing platform handling real-money payments, driven by payment webhooks, and hit by spiky on-sale traffic — and moved it off a single self-hosted server onto reproducible, self-healing Azure infrastructure. The goal was to make the platform scalable, auditable and safe to release, without ever putting payment flows at risk.
We defined the entire environment as code: modular Bicep templates provisioning Azure Container Apps, Azure SQL, Key Vault, Container Registry, storage and a full alerting stack. A gated Azure DevOps pipeline builds, runs database migrations safely at deploy time, rolls the container image and verifies the release with a health check. Observability runs on OpenTelemetry into Azure Monitor, with alert rules written around the things that actually matter — payments, payouts and webhooks.
63Digital didn't just host our platform — they turned our infrastructure into code, made every release safe, and set up alerting that tells us about a payment issue before anyone else notices. We deploy with confidence now.
A multi-stage approach.
Codify the infrastructure
We defined the whole environment as composable Bicep modules — Container Apps, Azure SQL, Key Vault, Container Registry, storage, managed identity and alerting — so the estate is reproducible, version-controlled and deletion-protected.
Safe, gated releases
We built an Azure DevOps pipeline that builds and tests, packages database changes into a self-contained migration bundle, and — behind a manual approval gate — runs those migrations against Azure SQL, rolls the container image, and verifies the deploy with a health probe.
Secrets and identity
Production secrets are resolved from Key Vault via a managed identity at startup, so no connection strings or keys are ever baked into an image — and the running app scales without touching a credential.
Payment-grade observability
We wired end-to-end OpenTelemetry into Azure Monitor and wrote alert rules around what matters — payment failures, payout-token renewals, webhook and email failures, background-job errors — with noise-suppression for internet scanner traffic and a synthetic uptime test.
What changed.
The whole estate, as code
Modular Bicep templates define the entire environment, so it can be rebuilt or reviewed like any other code — with a resource lock and a purge-protected Key Vault guarding it against accidents.
Releases that can't lose a migration
Every production deploy runs versioned database migrations from a self-contained bundle, updates the container image and self-verifies via a health check — behind a human approval gate.
Always-on, sized for launch night
The platform runs as an always-on container with no cold starts, scaled up to 2 vCPU and 4 GB ahead of a major launch, behind a custom domain with auto-renewing TLS. It deliberately stays on a single replica, so a busy on-sale can never send customers duplicate ticket emails.
Alerts that page on what matters
Alongside the usual error and latency checks, dedicated alert rules watch payments, payout-token renewals, webhooks, emails and background jobs — so the team hears about a payment problem before a customer does.
Outgrown a single server?
If your platform has become too important to run on hope and a single box, we'll move it onto codified, self-healing cloud infrastructure with releases you can trust.