Plan for losing a whole region, not one server
We design and test a recovery plan that survives the loss of a whole cloud region, then prove it with a timed failover drill.
Overview
On 1 to 2 March 2026, drone strikes hit AWS data centres in the UAE region me-central-1. Two of its three availability zones went down. Banks, payment providers and consumer apps had outages, and AWS told customers to move workloads to other regions. Teams that had relied on multi-AZ inside one region found it did not protect them.
We help you decide what must come back, how fast and where, then build it. That means a target region, replicated data, infrastructure you can rebuild from code, DNS you can switch, and a runbook your own engineers have used in a real drill. The work happens inside your own cloud accounts.
Signs you need this
- Your recovery plan assumes the region itself stays up
- Nobody can say how long a full rebuild in another region would take
- Container images, secrets or Terraform state live only in the region you would be leaving
- Your last failover test was a tabletop discussion, not a real switch
What you get
- Recovery objectives per system
- A short table, agreed with your business owners, giving the recovery time and data loss each system can tolerate and why.
- Region-loss dependency map
- Every hidden tie to the primary region: registries, KMS keys, secrets, state files, identity, third-party allowlists and hard-coded endpoints.
- Recovery environment in code
- Terraform or OpenTofu for the target region, with data replication set up and tested, so recovery is a pipeline run rather than manual clicking.
- Failover and failback runbook
- Step-by-step instructions with owners, commands and decision points, including how to return to the primary region afterwards.
- Timed drill and report
- A live failover exercise run with your team, with the measured time to recover and every problem found written down and fixed.
How the work runs
- 1
Impact and objectives
1 to 2 weeks
Interview system owners, list critical paths and agree recovery time and data loss targets for each tier.
- 2
Dependency review
1 to 2 weeks
Trace every regional dependency in your accounts and pick the target region with data residency rules in mind.
- 3
Build the recovery path
3 to 6 weeks
Write the infrastructure code, set up cross-region replication for databases and storage, and prepare DNS and traffic switching.
- 4
Drill and harden
1 to 2 weeks
Run a planned failover with your engineers, measure it, fix what broke and schedule the next drill.
Timings depend on the size of your estate. The assessment fixes them before any build starts.
What you can check afterwards
- A failover to a second region that you have watched complete, with the measured time written down
- A runbook your own on-shift engineer can follow without us on the call
- No production dependency left that exists only in the primary region
- Recovery objectives your board and auditors can read and sign off
Tools we work with
- AWS
- Azure
- Google Cloud
- Terraform
- OpenTofu
- AWS Elastic Disaster Recovery
- Amazon Route 53
- Azure Site Recovery
- Amazon Aurora Global Database
- Velero
- Kubernetes
Questions buyers ask
- Will the drill cause downtime for our customers?
- We start with a drill against a copy of production or a quiet window, so customer impact is planned and agreed in advance. A full production failover only happens when you choose to run one.
- What access do you need?
- Read access to start, then a role in your own cloud accounts scoped to the recovery work. Everything we build stays in your accounts and your Git repositories.
- Does the second region have to be outside the UAE?
- Not always. If data must stay in-country, we look at a second provider's UAE region instead, which our multi-region service covers in more depth.
- What happens after the project?
- You keep the code, the runbook and a drill schedule. We can come back for the next drill, but the plan is written so your team can run it alone.
Talk to us about disaster recovery and region-loss readiness.
Tell us what you run and where. We reply within one business day with questions or a quote.