Facebook’s outages in September had a significant impact on users.
The downtime experienced by Facebook in September dropped its reliability score from 100 in August to 97 percent, putting it below both Tumblr and Twitter. The biggest social media company also had the slowest-performing API, which, on average, was 2.4 times slower than Twitter. These same results were found across all cloud services tested in Europe, Asia and US.
One finding shows that the choice of cloud provider had a measurable impact on social media API performance, equal to or greater than the impact of a user’s geographic location. While Google Cloud offers the best performance across all regions, Microsoft Azure was shown to have reliability and performance problems impacting every social network across every region. Latency for each platform also varied significantly between cloud providers, from 342ms for Twitter on AWS to 1240ms for Facebook on Azure. This is one of the main reason why developers should take these differences and the location of their users into consideration when designing applications and choosing cloud providers.
“API performance issues can have enormous ramifications, but many developers and IT managers are monitoring them incorrectly or not at all,” said David O’Neill, CEO and co-founder of APImetrics. “Poor API performance is a leading indicator of potential application issues, so our goal is to give organizations actionable data that informs decisions about their applications, and helps them troubleshoot issues before users notice them.”