re: Facebook Malware Crackdown Caused App Outage
Automated operations in some cases are going to contain their own problems, rather than allowing us to move smoothly into the software-defined data center. The larger the scale, as Amazon learned to its regret on Easter weekend 2011, the greater the mishap when an automated process goes awry. Unplugging one network and replugging the traffic into an alternative and under-capacity network set off "a re-mirroring storm." We need to get smarter about anticipating what can go wrong with automated systems and build in some protection logic. Much of this learning, however, will occur the hard way.