APImetrics – API Metrics 2016 By The Numbers [INFOGRAPHIC]

After conducting the largest analysis of API performance ever undertaken, we’ve reached the same conclusion as we did when we first started with APImetrics. Half of the APIs we survey STILL fall below expected quality standards.

As Kin Lane, the API Evangelist says, “You wouldn’t operate an API without having the right data regarding it’s availability and stability—the same is true for the API industry, and the folks at APImetrics have those numbers.”

In our original study, we were only able to focus on conventional error or success codes like HTTP 200. We didn’t have the sophistication to understand more of the context of our findings.

For example, we were able to see that over half of all the APIs we monitored had downtime during the week. But weren’t able to conclude if that would impact users negatively. With our advanced monitoring tools like CASC , we can draw more detailed conclusions around the impact of API performance issues on users and providers.

We still see that 50% of the APIs we monitor fail to meet our qualification for Good performance. That potentially leads to more than 200 wasted engineer-years of effort investigating unexpected performance problems.

The Infographic we developed shows us a snapshot of the performance of the API calls we monitored over the course of 2016. In addition to seeing the performance problems, we were able to see regional and cloud-specific variations in API performance that should be considered when architecting an API- or Micro-service-based solution.

“What shocked us is that the results were far worse than we expected,” O’Neill said. “Not only did over half of all APIs have issues, but those issues caused the APIs to fall well below best practice standards. In other words, those API problems do affect the consumer – and can be deadly to a company's bottom line.”

Try us out

We also were able to show that the location and cloud provider can significantly impact performance in ways that can’t purely be explained by geography. Calls made from Brazil took 3 times longer than calls made from the US West Coast. Calls made from Asia-Pacific, almost twice as far away, took just twice as long.

But, that’s far too simplistic for the results we saw.  Calls made from Google’s Cloud Host in Taiwan took just 22% more time to read most APIs than similar calls made in the Western US. Calls from Hong Kong could take twice as long.

In our next report we’ll go into more detail concerning each of the regions we’re monitoring – and what it can tell you about where to house your API servers to optimize performance.

For now, feel free to download our Infographic and do not hesitate to contact us regarding any questions you may have.

Fill in the form to download the Infographic for free

By | 2017-10-06T02:45:02+00:00 April 19th, 2017|API Update, Case Studies, Press Releases, Sectors, Uncategorized|0 Comments

Leave A Comment