Daniel Fridkin · Home
עברית
עברית
Back to projects

Platform Architecture / Cloud Migration / HA

OCI Frankfurt Multi-AD Production Migration

High-availability production migration program moving critical e-commerce workloads from single-AD constraints toward OCI Frankfurt across three availability domains.

In progress

Migration target

From delivery controls to resilient workloads

A conceptual view of the planned Frankfurt migration. Multi-AD placement and rollback are part of the target design; the migration is in progress.

  1. Provision & govern

    Networking, compartments and controlled delivery.

    • Terraform
    • Argo CD

  2. Place workloads

    Kubernetes across three Frankfurt availability domains.

    • OCI
    • Kubernetes

  3. Plan stateful services

    Failure-domain placement, observability and recovery criteria.

    • Redis
    • Elasticsearch
    • MariaDB
    • ProxySQL

The challenge

The existing production footprint is exposed to regional and connectivity failure modes, while critical retail workloads require stronger availability ahead of peak traffic periods.

The approach

Designed a phased migration model across three Frankfurt availability domains covering Kubernetes, networking, compartments, observability, stateful services, deployment controls, and rollback/failure-domain planning.

My role

Lead architecture and execution planning across platform, networking, Kubernetes, stateful workload placement, observability, security, and operational readiness.

Engineering decisions

Plan around failure domains

Move beyond a single-AD footprint; plan both stateless and stateful placement across three availability domains.

Make the cutover reversible

Use migration phases with explicit observability and rollback criteria alongside the Cloudflare cutover strategy.

Scope & delivery

  • Multi-AD Kubernetes placement and failure-domain-aware workload design.
  • Migration planning for roughly 270+ pods / 300+ containers.
  • Stateful service planning across Redis Sentinel, Elasticsearch, and MySQL/MariaDB with ProxySQL patterns.
  • Cloudflare routing/security integration and controlled cutover strategy.
  • Terraform-driven networking, subnet, compartment, and security-group redesign.
  • Observability and rollback criteria designed into migration phases.
Next projectGrafana Alloy Fleet Manager