ra2 studio - Fotolia
Amazon Web Services will reboot a portion of its server fleet over the next two weeks.
"We have built the capability to live-update the vast majority of our fleet; however, we have not yet enabled this capability on some of our older hardware," the advisory read. "This older hardware is what's being rebooted."
Instances that are rebooted will return to normal operation after the reboot, and instance configuration and data will be retained, including data on local hard drives, the advisory said. Users will not be able to stop and start or re-launch instances to avoid the update if their hardware is affected. Startup procedures that aren't automated during the instance boot process will need to be re-run.
"It does not at all scare me," said Kevin Felichko, CTO of PropertyRoom.com, an online auction company based in Frederick, Md. "We architected our infrastructure to handle failure as if it were a normal thing. If we receive notice [that it will affect our servers], we will simply mark the maintenance window on our calendars and verify all is well."
John TreadwaySenior Vice President, Cloud Technology Partners
Undoubtedly, the reboot will impact some users, according to John Treadway, senior vice president with Cloud Technology Partners, based in Boston. With proper planning and operational controls, the reboot should not cause customers any real difficulty, Treadway added.
"The bottom line is that this is really not a big deal and is actually evidence that AWS is on top of their infrastructure security," Treadway said. "Of course, if you are not paying attention and properly managing your AWS environment, the impact may be painful. But really, whose fault is that? Not Amazon's."
The AWS advisory also contains a frequently asked questions session that hints that the reboot could affect some secondary services. Users will be notified if instances that underpin WorkSpaces, Relational Database Service (RDS), ElastiCache and RedShift are affected.
AWS has also pledged not to reboot instances in multiple regions at the same time, especially with RDS instances.