Guillermo Rauch

Vercel's Dubai region (dxb1) on AWS me-central-1 experienced an incident where availability zone mec1-az2 was impacted. The primary traffic ingress AZ remained unaffected, and Fluid Functions were unprotected due to automatic multi-AZ deployment and load balancing. Vercel's infrastructure supports automatic traffic rerouting and multi-region failover, ensuring continued access to critical services during regional disruptions.

Last year we announced the Vercel Dubai region (𝚍𝚡𝚋𝟷) on AWS 𝚖𝚎-𝚌𝚎𝚗𝚝𝚛𝚊𝚕-𝟷.

A region is made up of multiple availability zones (AZs). The AWS availability zone 𝚖𝚎𝚌𝟷-𝚊𝚣𝟸 just got 💥 bombed.

Our primary traffic ingress AZ has been unaffected. Fluid functions are also unaffected, because they automatically deploy to multiple AZs and load balance around them.

AZs are basically “sub-regions”. They’re designed to have independent power supply, cooling, networking, physical security, fire suppression, and logistical operations.

If Dubai’s region got seriously impacted, Vercel automatically reroutes traffic. Fluid functions can deploy to a backup region for automatic failover. So you get multi-AZ and multi-region. In a situation like this, this means citizens being able to access critical information, news, emergency services, and so on.

Here’s hoping the situation normalizes as soon as possible and peace prevails 🤞

<p>Last year we announced the Vercel Dubai region (𝚍𝚡𝚋𝟷) on AWS 𝚖𝚎-𝚌𝚎𝚗𝚝𝚛𝚊𝚕-𝟷.

A region is made up of multiple availability zones (AZs). The AWS availability zone 𝚖𝚎𝚌𝟷-𝚊𝚣𝟸 just got 💥 bombed.

Our primary traffic ingress AZ has been unaffected. Fluid functions are also unaffected, because they automatically deploy to multiple AZs and load balance around them.

AZs are basically “sub-regions”. They’re designed to have independent power supply, cooling, networking, physical security, fire suppression, and logistical operations.

If Dubai’s region got seriously impacted, Vercel automatically reroutes traffic. Fluid functions can deploy to a backup region for automatic failover. So you get multi-AZ and multi-region. In a situation like this, this means citizens being able to access critical information, news, emergency services, and so on.

Here’s hoping the situation normalizes as soon as possible and peace prevails 🤞</p>

Vercel is experiencing a multi-AZ outage affecting its Dubai (dxb1) region, described as potentially the first of its kind caused by a non-software root cause. Traffic is being rerouted to bom1 (Mumbai), and Dubai is being excluded from all Function and Routing Middleware deployments. Node.js-based middleware exclusions are complete, with edge runtime exclusions in progress. Users experiencing issues are advised to switch to the Node.js runtime. Vercel is also reinforcing emergency failover measures for faster disaster recovery.

Update on 𝚍𝚡𝚋𝟷. The situation has worsened and it’s progressed into a full n blown, multi-AZ outage.

(Possibly first ever of its kind ever from a non-software root cause)

Many workloads on Vercel are global in nature. Routing Middleware for instance automatically deploys compute to every region.

In addition to rerouting traffic successfully to 𝚋𝚘𝚖𝟷, we’re also excluding Dubai from all Function and Routing Middleware creations. This has been completed for Node.js-based middleware, and our teams are working to extend this to edge runtimes. (If you’re seeing issues with your deployments, consider switching to our recommended Node.js runtime.)

As we face an uncertain global landscape, we will continue to reinforce our emergency failover measures, for even faster disaster recovery.

<p>Update on 𝚍𝚡𝚋𝟷. The situation has worsened and it’s progressed into a full n blown, multi-AZ outage.

(Possibly first ever of its kind ever from a non-software root cause)

Many workloads on Vercel are global in nature. Routing Middleware for instance automatically deploys compute to every region.

In addition to rerouting traffic successfully to 𝚋𝚘𝚖𝟷, we’re also excluding Dubai from all Function and Routing Middleware creations. This has been completed for Node.js-based middleware, and our teams are working to extend this to edge runtimes. (If you’re seeing issues with your deployments, consider switching to our recommended Node.js runtime.)

As we face an uncertain global landscape, we will continue to reinforce our emergency failover measures, for even faster disaster recovery.</p>