Gmail suffered a prolonged service outage yesterday, with tons of users reporting that they weren’t getting any email, and that sent emails were getting delayed — sometimes for over two hours. Google has since fixed the issue, and they announced today that the issue was caused by a “dual network failure,” which “is a very rare event.”
A dual network failure is when two separate, redundant network paths both stop working at the same time, thus leading to problems with emails from being sent through. Most network systems consist of a backup system that takes over if the primary network path breaks down. However, this time around, Google says that both systems failed at the same time.
This, in turn, “reduced Gmail’s capacity to deliver messages to users,” meaning that some users weren’t able to send and receive emails while others were still able to. Google says the problem took about 10 hours to fully fix, but were able to start clearing up email blockage after about seven hours.
However, the problem affected more users than Google expected. At first, the company reported that “less than an estimated 0.024% of the Gmail user base” was affected, but we received many emails and comments from readers pointing to issues on their end, as well as co-workers whom they work with in the office, leading to the assumption that it definitely affect more than 0.024% of Gmail users. Even a few of us here at GottaBeMobile were experiencing issues with Gmail.
However, Google says that only around 1.5% of Gmail emails are delayed more than two hours, with 29% seeing an average delay of around 2.6 seconds, with 71% of emails seeing no delay. 1.5% is still a fairly low number of Gmail users, but seeing as how the service has tens of millions of users, a single percentage point is quite a lot, and it’s definitely more than Google’s initial estimate of 0.024%
This certainly isn’t the first time that Gmail has gone down, but it’s also not a common occurrence, especially when it’s down for over the entire work day (and on a Monday — ouch!). The timing certainly could’ve been a little better, but network systems sadly don’t take human emotions into consideration.