Tag Archives: resilience planning

AWS, Middle East Escalation and a Very Real Reminder That the Internet Is Physical

Over the weekend something unusual happened.

The AWS Health Dashboard lit up with warnings across the Middle East regions, while at the very same time global news outlets were reporting escalating military action across the Gulf.

If you run infrastructure in the cloud, or even if you just assume “the cloud” is always there, this was a sobering moment.

Let’s unpack what actually happened.


What Amazon Web Services Said

According to the official AWS Service Health Dashboard:

“Objects struck the data centre, creating sparks and fire.”

That is not typical outage language.

AWS reported that two Availability Zones in the ME-CENTRAL-1 region were impaired due to a localized power issue. Power was shut off while emergency services responded, and recovery would require:

  • Repair of facilities
  • Restoration of cooling systems
  • Restoration of power systems
  • Coordination with local authorities
  • Safety assessments before re-energising the site

Customers were strongly advised to fail over to alternate regions, ideally in Europe.

This was not a minor API hiccup. EC2, S3, DynamoDB, the AWS Management Console and dozens of other services experienced elevated error rates.

The cloud, quite literally, caught fire.

Source: AWS Service Health Dashboard


At The Same Time… The World Was On Edge

Simultaneously, live coverage from BBC News reported dramatic escalation in the region:

  • Iranian strikes on a major gas plant in Qatar
  • A refinery fire in Saudi Arabia
  • Drones intercepted over Cyprus heading towards a UK base
  • Escalating conflict involving Israel, Lebanon and Iran
  • Reports that Iran’s Supreme Leader had been killed in US and Israeli strikes

Gas prices reportedly spiked sharply on international markets.

Source: BBC Live Coverage

AWS has not explicitly linked its data centre incident to military activity.

But when you read phrases like “objects struck the data centre” in the same time window as confirmed missile and drone activity across the Gulf, the coincidence is difficult to ignore.


The Myth of the Abstract Cloud

We talk about:

  • Serverless
  • Containers
  • Regions
  • Availability Zones
  • Multi-AZ architecture

All wonderfully abstract.

But this weekend was a reminder that the cloud is:

  • Concrete buildings
  • Power substations
  • Cooling plants
  • Diesel generators
  • Fibre routes
  • Security perimeters

Remove electricity and you remove the cloud.

Damage cooling systems and you shut down racks.

If local authorities tell you to keep power off, your “infinite scalability” suddenly looks rather finite.


Why This Matters To You

If you deploy only in one region, you are accepting regional geopolitical risk whether you realise it or not.

AWS always recommends multi-AZ design. Many organisations stop there.

But this incident affected more than one Availability Zone in the same region. That is the critical detail.

Multi-region redundancy is no longer theoretical resilience planning. It is operational reality.

If your backups sit in the same geography as your primary systems, that is not true disaster recovery.


The Bigger Lesson

We spend huge amounts of time worrying about:

  • Cyber attacks
  • Zero day exploits
  • Ransomware
  • Misconfigured S3 buckets

Yet physical risk is often treated as someone else’s problem.

This event shows that geopolitical instability can ripple directly into cloud availability.

Cloud providers are extraordinary at redundancy. But they are not immune to real world events.

When missiles fly and power grids are shut down, even hyperscale infrastructure feels it.


Final Thoughts From The Server Rack

I have long argued that we live in a world where digital and physical are inseparable.

This weekend was a perfect example.

A regional conflict.
Energy infrastructure under threat.
Data centres hit.
Gas markets spike.
APIs fail.

The internet is not floating in the ether. It is bolted to the floor.

If you are running production workloads, ask yourself one simple question:

If my region goes dark for 24 hours, what happens next?

If the answer is panic, then this weekend was your warning shot.

As ever, the smartest architecture is not the cleverest. It is the most resilient.

And resilience, increasingly, means geography.