Results at a glance
- 18+servers provisioned across availability zones
- IPsecencrypted tunnels for all inter-zone traffic
- IP setsand application rules behind every edge firewall policy
- Cross-zonereplication for disaster recovery
The brief
The platform needed to survive the loss of a whole zone, keep its management and service traffic private, and still be reachable by customers. The environment was a VMware Cloud Director tenant, so everything was built in vCD: VMs, vApps, storage and edge networking.
What I built
Compute and storage
I created the VMs and vApps, assigned storage pools, and installed and configured the operating systems and production services on every machine.
Network and security
- Edge gateways per zone, with firewall policies built from IP sets and application rules instead of one-off addresses. That makes rules readable and easier to audit.
- NAT rules to publish only the services that must be public.
- IPsec VPN tunnels between zones, so replication and management traffic never cross the internet in clear text.
- Application security groups and routing for the whole topology.
Disaster recovery
Cross-zone replication keeps a recent copy of critical workloads in another zone, so the service can be restarted there if a zone fails.
Decisions and trade-offs
- Named objects over raw IPs. IP sets and groups take longer to set up but make every later change and audit faster and safer.
- Encrypt east–west traffic too. Inter-zone traffic runs over provider infrastructure, so I treated it as untrusted and put it in IPsec.
Outcome
The platform runs across multiple zones with a documented, auditable network design, private inter-zone links and a recovery path if a zone is lost.
