API.expert Weekly API Analysis: Hey Foursquare!

This Week’s API Highlights

• Twitter in the Social Networks API category takes the overall title this week with a CASC score of 9.74.
• Department of Justice in the US Government API category takes the overall title again with a median latency of 47 ms, the same as last week.
• General Services Administration in the US Government API category takes the overall bottom spot this week with a CASC score of 6.00.
• We take a good, hard look at what went wrong with Foursquare this week.

Something of interest

Foursquare appears in two of our API categories, Search and Social Networks. Unfortunately this week, it finds itself in the Amber Zone in both and propping up each table. What went wrong? In both API categories, Foursquare has a pass rate below 97% and ~5% of outliers, which aren't ideal figures. Looking at the Insights report for the Foursquare tests in each API category, we can see the problem immediately – major outage on 8 April. The Foursquare User Profile Fetch API in Social Networks (below) runs at a higher cadence than the Foursquare Search test in the Search category, so we see more a higher number of failures (the overall failure rate is similar), which separate out into two distinct groups, one with calls with a latency of about 5000 ms and one with calls with a latency of about 10,000 ms. By actively monitoring your APIs, you can understand exactly when they are going wrong and exactly how they are going wrong and tease out subtle patterns in what is going on with the API. The vast majority of failures were HTTP status code 500 responses, an Internal Server Error with the message:

{ "meta": { "errorType": "server_error", "code": 500, "errorDetail": "Foursquare servers are experiencing problems. Please retry and check status.foursquare.com for updates." }, "response": {} }

We also saw 504 Gateway Time-error during the outage and a handful of 503 No healthy backend errors outside it over the last week. The outages started 19:58 UTC on 8 April for both endpoints and ended to 1:07 UTC on 9 April for the User Profile Fetch and at 0:29 UTC on 9 April for Search, about five hours. This was a pretty major outage for a big service like Foursquare – likely with significant impact on user experience. It’s through active monitoring that you know about these problems as soon as they arise and move to fix them immediately.

API Analysis: Tops in Overall Quality

Week Ending 12 April 2021

Category

Organization

CASC score

Corporate Infrastructure

Slack

9.58

COVID-19

WHO