Picking up right where you left off in the event of a failure will now be possible for those running Elastic Computing Cloud (EC2) instances on AWS.
Previously, if customers wanted to set an EC2 instance to recover automatically, they could only do so by setting up an alarm in Amazon CloudWatch.
Now though, if an underlying hardware problem comes along, EC2 instances will be automatically recovered on Amazon’s cloud computing service where they’ll receive a new instance ID, private IP addresses, public IPv4 IP address, Elastic IP addresses and all instance metadata. Unfortunately though, data in memory will still be lost in the event of a failure.
Automatic recovery of EC2 instances
While automatic recovery of EC2 instances will likely be a big deal for businesses that run their workloads in the cloud, AWS only released a brief announcement regarding the new feature.
While the announcement also contained a link to the company's documentation on how to recover instances, it didn’t mention anything about recovery points for data on disks or the time required for an auto-recovered instance to resume operations.
EC2 customers will also be able to disable the new auto-recovery feature. For most customers this probably isn’t a good idea but AWS points out in its documentation that those running instances in placement groups might want to disable the feature so that their instances are restored to the same placement group.
Still though, hardware faults in a server could indicate that there is a problem in a server rack or even an entire row which is why having an instance auto recover to another availability zone or region makes sense.
Via The Register
Generative AI has a long way to go as siloed data and abuse of its capacity remain a downside – but it does change the game for security teams
Diamond set to become mainstream coolant for AI GPU servers as world’s best thermal conductor promises 25% better overclocking, and 'double performance per watt'
