Amazon data centres face months
- Amazon said AWS regions in Bahrain and the UAE will need several more months of repairs after Iranian drone strikes damaged facilities in March. - AWS says UAE workloads remain only partly functional, Bahrain is still unavailable, and billing is paused while customers move apps to other regions. - The bigger point is simple: cloud looks virtual, but hyperscale computing still depends on vulnerable buildings, power, cooling, and local access.
Cloud outages usually sound abstract — packets, regions, failover plans, maybe a dashboard full of red dots. But this one got physical fast. Amazon now says its AWS regions in Bahrain and the United Arab Emirates will take several more months to recover after Iranian drone strikes in March damaged facilities and nearby infrastructure. That matters because these are not edge caches or backup closets. They are full cloud regions that businesses use to run real applications, store data, and meet local residency rules. (arstechnica.com) ### What actually got hit? AWS has said two facilities in the UAE were directly struck, while a strike near one Bahrain facility caused physical damage there too. The public service notices describe the UAE region, ME-CENTRAL-1, as unable to reliably support customer applications, and the Bahrain region, ME-SOUTH-1, as currently u(arstechnica.com). Here, the buildings and supporting systems took the hit. (health.aws.amazon.com) ### Why does “several months” matter so much? Because the timeline tells you this is not a quick swap of failed servers. Data centers depend on power gear, cooling equipment, networking hardware, fuel logistics, physical security, and safe access for crews. If any of those are damaged — or if the area stays unstable — repairs drag out. Ars Technica notes that the full disruption could stretch(health.aws.amazon.com) cloud terms. (arstechnica.com) ### Why can’t Amazon just fail over everything? Some workloads can move. AWS is telling customers to migrate accessible resources to other regions and restore inaccessible ones from remote backups. But the catch is that failover only works if customers built for it in advance. A lot of companies choose a local region for latency, (arstechnica.com)ally buy it for you. (health.aws.amazon.com) ### Who feels this first? Companies that pinned production systems to those Middle East regions. Earlier reporting tied the disruption to enterprise and consumer services in the Gulf, including apps and corporate workloads that depend on AWS being nearby and continuously available. Even when an app comes back, performance, data sync, and operating costs can all get worse if traffic has to run from Europe, Asia, or the US instead. (restofworld.org) ### Why pause billing? Amazon says relevant billing operations are suspended while it restores normal operations in the affected regions. That is partly customer relief, but it is also an admission that these regions are not delivering their normal service level. If your compute region is down or unreliable for months, charging as if nothing happened would be hard to defend. (sta([restofworld.org)-off, or a bigger warning? It looks like a bigger warning. The cloud is sold as weightless and everywhere, but it still lives in specific places with fences, substations, diesel tanks, chillers, and fiber routes. Those places can be hit by war, weather, sabotage, or grid failures. The more countries push for local AI and local data storage, the more valuable — and exposed — those physical sites become. (arstechnica.com) ### So what changes now? Expect more pressure on multiregion design, more scrutiny of where cloud regions are built, and more awkward conversations about whether civilian tech infrastructure now sits inside military risk maps. Basically, this story is not just about Amazon fixing damaged sites. It is about the industry being reminded that “the cloud” is still a set of buildings on the ground. (arstechnica.com)
