The Cloud Has Geography Again

Amazon Web Services told customers on 15 September, via the Health Dashboard, that it cannot restore access to resources and data hosted exclusively in the Middle East (Bahrain) Region, me-south-1. The damage “spanned multiple Availability Zones and exceeded what our regional and multi-AZ services are designed to withstand.” After assessment, exclusive-hosted copies in that region will not come back. A parallel post said the same for one UAE Availability Zone, mec1-az2, in me-central-1. Recovery work continues on mec1-az1 and mec1-az3. Most customers, AWS said, had already restarted in other regions from backups or remaining copies. That is the wire. Reuters, CNBC, the WSJ, and The Register all saw the same language. It is Amazon’s first public status update on the sites since April.

The residual is not that Amazon lost “some data.” It is that the abstraction failed in the way architects were told it would not. The first war to hit the cloud was a March outage story. Tuesday is a permanence story.

Multi-AZ is not a second country

An availability zone is one or more data centers in a region, isolated enough to survive a building fire or a utility cut. It is not isolated enough to survive a campaign that can hit more than one Gulf campus. The Register’s reconstruction, matching AWS’s own March–April guidance, is sequential: a first Bahrain zone damaged in March, a second disrupted in April, then the region treated as unavailable. AWS’s September sentence is the design limit, not a rumor: multi-AZ services were not built for that blast radius.

What is gone is data that lived only there. Customers who followed the March advice — migrate, restore from remote backups, point traffic away — are the “most” in the dashboard. The WSJ’s reading is the remainder: exclusive Bahrain and UAE copies may be gone forever. IRGC claims of a July cruise-missile “destruction” of Bahrain sites, and satellite images it released of Zallaq and Askar, are Iranian statements. AWS did not re-litigate them on Tuesday. It said the region cannot be restored for unmigrated resources, with a further Bahrain customer update expected in early 2027 and UAE restoration notes “in the coming months.”

Virginia plus Ohio is still one polity

Enterprise disaster recovery still often means two U.S. regions, or two zones in one Gulf region for data-residency. That is how an investor would not diversify banks. Kinetic risk correlates at the theater: Bahrain and the UAE were both in the same war. A container that can leave a pad is one response to geography. Another is a second copy in a second legal and military system, not a second suburb.

The dashboard also said AWS had “exhausted every option” for data not migrated before the Bahrain region went dark. That is an operational close, not a legal finding of who owes whom. Billing in the affected regions had already been suspended for months; Careem’s overnight hop to other servers, reported in the spring, is the success case. Markets have priced Gulf peace that did not reopen physical systems. Cloud DR that assumed the building would still exist is the same class of error.

Logical redundancy failed where the buildings shared a war.

What still mis-prices is treating us-east-1 plus us-east-2 as a geopolitics policy, or treating Tuesday as proof that every replica in the Gulf is gone. The test is architectural and dated: whether 2027 DR runbooks require a copy outside the theater — and whether AWS’s early-2027 Bahrain note is a rebuild plan or a write-off. Coordinates were always there. War made them the SLA.