APImetrics has announced the launch of the APImetrics API Performance Test Solution, a new kind of end-to-end API testing and monitoring service that tests an API’s functionality and speed of response using complex, authenticated API calls. Most API testing services typically check to see if an API is online by pinging the API address and then reporting the latency of the ping. Using API calls allows the APImetrics service to provide comprehensive monitoring and analysis of the actual performance of APIs.
APImetrics also announced the results of API monitoring and testing for the 30 days preceding November 13, 2013. The results showed that APIs from leading API providers such as Amazon, Google, LinkedIn, Intel and Instagram reported outages at some point during the time period.
The proliferation of mobile devices, the rapidly growing Internet of Things (IoT), Open Data initiatives, and many other technological trends have led to an exponential increase in the number of actively used APIs. Today API providers such as Twitter, Google, Facebook, Netflix and many others handle billions of API calls (View the infographic in this post to see who is a member of the API Billionaires Club). It will not be long before the number of APIs available reaches a million making it all the more important that all active APIs are reliable, monitored and that the actual performance of APIs are routinely tested. David O’Neill, CEO APImetrics, explained to ProgrammableWeb the key to APImetrics’ ability to scale:
“The shear number of APIs is pretty daunting. Key to being able to handle large numbers is making it as easy as possible to create, share and configure tests. This meant having a system for generically handling security tokens. We support OAuth 1.0 and 2.0 in our product, but before we started we didn’t realize that the implementation of OAuth 1.0 in Flickr bears very little resemblance to the one used by Twitter. Likewise, where Facebook’s OAuth 2.0 easily renews tokens, Salesforce’s version doesn’t. Handling authentication and tokens generically is key to our ability to scale.”
The APImetrics platform features an easy to use test editor and token handling system so that developers, enterprises, API owners and even non-developers with some technical knowledge can use the service. Premium and enterprise clients can keep their tests private, however, there is also a community for sharing test and performance data that anyone can contribute to. One example of a publicly shared test call is IP Utilities – Get my IP which was provided by Mashape.
Mashape has shared additional API test calls with the APImetrics community such as Yoda Speak, Word Cloud Maker and Currency Converter. APImetrics also has an APImetrics Public Report that shows the hourly, daily and weekly status of widely used APIs such as Amazon OAuth 2.0 dialog, GitHub OAuth 2.0 dialog, Facebook Share, Tumblr Share, Twitter Share and many others.
APImetrics is working on setting up standardized metrics to determine an API’s performance and how to rank APIs. David O’Neill told ProgrammableWeb that APImetrics is looking for three things in the data: failure rates, latency figures and variability in the latency numbers. O’Neill also mentions that APImetrics is looking at various factors from a quality perspective. “This is a hard problem as it involves considering how to rank APIs against multiple factors like reliability and latency,” he says. Kin Lane, API Evangelist, comments about APImetrics in the press release:
“Nearly every major website today either provides data to others through an API or consumes an API to deliver functionality to their users. Many do both and use multiple APIs, all of which makes identifying problems with APIs critical. APImetrics is the first service I’ve seen that measures the actual performance of APIs in a way that matches how individuals and companies experience APIs.”
For more information about the APImetrics end-to-end API testing and monitoring service, visit apimetrics.io.