amazon_aws_logo.jpgSo, you move your company's apps, services or data to the cloud, and think everything will roll on smoothly, forever. Wrong! Even Amazon's systems need the odd update that requires a reboot within its mighty data centers.

A Firm Boot

Deep within Amazon's mega data centers, the update process has already started with an unspecified number of EC2-running systems requiring a physical reboot to complete the installation of important security patches and other updates, with the update task potentially going on over the next week.

Users have been pre-notified about the updates, along the lines of "instance (i-exxbbbxx) would require a reboot on xx December" and have the option to run them at their own preferred time, rather than waiting for Amazon's own equivalent of a man turning up to press the button. Note, some users are doing their own reboot and still seeing the "upcoming maintenance reboot" message; this is due to lag on Amazon's side.

Despite the advance notice, there has been the usual gnashing of teeth on Twitter from some users, even though hard reboots sometimes happen due to software or hardware issues. Those unable to clone their services over to another Amazon center may see some brief downtime -- just a few minutes according to Amazon.

There are no reports of any wider outages, and Amazon's trouble board is green, so this process seems to be going smoothly so far. See Amazon's own help page for detailed information.

The Big Delete Button

Things may get a bit more heated tonight; one tweeter mentions "my amazon aws rep just called me, telling me they are rebooting entire west coast tonight, any non-EBS backed instances will be lost" -- if that's true, lots of users may start complaining.

Elsewhere in Amazon Land, those unconcerned by these updates seem to be celebrating the addition of a new feature. Multi-object delete allows the removal of up to 1,000 objects from an Amazon S3 storage bucket in one request -- a great improvement on past options.