En Irlande, un orage perturbe le datacenter d'Amazon

Le cloud computing n'est pas à l'abri ... de la foudre. La vraie, celle qui s'est abattue par exemple, dimanche dernier sur un transformateur près de Dublin. Résultat : le datacenter d'Amazon  s'est retrouvé bloqué entraînant une interruption de service (des serveurs EBS, Elastic block store) comprise entre 24 et 48 heures. Il a fallu trois heures pour la première opération de récupération de données. Douze heures après l'interruption, un quart des données étaient encore inaccessibles.

Reste à connaître la réaction des clients, leur facilité pour récupérer leurs données, leur opinion sur la qualité de l'information fournie par Amazon, la disponibilité qui leur était garantie (99,95%), la responsabilité réelle d'Amazon.

D'autres sociétés, dont Microsoft, seraient également touchées, toujours en Irlande, selon un de nos confrères spécialisé dans la bourse.



A litre, sur le site d'Amazon,  http://status.aws.amazon.com/   les informations données par la société (dans la journée du 7), dont voici quelques extraits :



RESOLVED] Slow Propagation Times

11:29 AM PDT We are investigating slow propagation of DNS edits to the Route 53 DNS servers. This does not impact queries to existing DNS records.

12:51 PM PDT Change propagation is back to normal. From approximately 10:48AM to 12:37PM PDT changes were delayed to a single edge location. Queries were not impacted during this time

 

[RESOLVED] VPN connectivity issues

12:09 PM PDT We are currently investigating connectivity issues to one of our redundant VPN Gateways in the EU-WEST-1 region. Each VPN Connection is provisioned as a pair of redundant VPN tunnels to provide customers with the ability to configure a VPN tunnel to two VPN Gateway endpoints. The redundant VPN Gateway is not affected and VPN Connections with both tunnels configured are operating normally.

1:11 PM PDT We can confirm that between 10:47 AM PDT and 11:34 AM PDT, one of our VPN Gateway devices in the EU-WEST-1 region was not available. Customers with both tunnels configured on their customer gateway devices were not affected. The issue with the VPN Gateway device has been resolved, and both tunnels for preexisting VPN Connections are now available. At this time, we are investigating increased latencies for Create, Terminate, and Attach operations for Virtual Gateways and VPN Connections.

2:14 PM PDT Latencies for Create, Terminate, and Attach operations for Virtual Gateways and VPN Connections have returned to normal. The service is operating normally.

 

RESOLVED] Delays provisioning stacks in EU-WEST-1 region

11:47 AM PDT We are currently investigating delays in provisioning CloudFormation stacks in the EU-West-1 region.

12:50 PM PDT Describe calls giving visibility into existing stacks continue to function, however, we are currently unable to provision new CloudFormation stacks in the EU-West-1 region. We are continuing to work on the issue.

1:18 PM PDT CloudFormation is now creating new stacks and the backlog of stack creations is draining.

2:27 PM PDT The backlog in creating new CloudFormation stacks has been cleared. The service is operating normally.