APImetrics Releases Largest Ever Cloud API Performance Report

Cloud performance is improving but an over reliance on a single data center location presents a serious risk to quality and performance

APImetrics Releases Largest Ever Cloud API Performance Report

The Definitive Annual Industry Report from the team, ecosystem, and data behind APImetrics.

APImetrics, the industry authority on API quality and performance, released the 2022 Cloud Performance Report, the company announced today.

“APIs are an essential part of the fabric of everyday life, and APImetrics is focused on how good the quality of those APIs are and the clouds that support them,” said David O’Neill, CEO of APImetrics. “People are missing how important the cloud is to the good performance and operation of APIs and it’s costing time, money and customer satisfaction.”

The report is based on the unique and ever-expanding API dataset generated by the APImetrics platform to establish an unbiased, industry-wide baseline for API quality scoring. It includes data from over a billion API calls to more than 8,400 different API endpoints from 70 geographically diverse cloud data centers across AWS, Azure, Google, and IBM.

The report focuses on data from leading API services, including those from prominent corporate infrastructure providers, financial services institutions, social networks, and search engines.

“Having significant amounts of the world’s cloud services reliant on one location from one provider is a huge cyber security risk”

— David O’Neill

Marketing Technology News: Loyalty Simplified: Smart Ways to Boost Grocery Loyalty With Technology

“APImetrics have been running standardized testing against some of the world’s most critical APIs for a long time now and have built up a massive and unique dataset that gives us insight not just into the functionality of those APIs but also the cloud services that they rely on,” O’Neill said.

Among the key findings is that for the fourth year in a row, AWS US-East (N. Virginia) was the fastest cloud location for Time to Connect with an average time of 1.23 ms, demonstrating that many solutions continue to be hosted here.

“Having significant amounts of the world’s cloud services reliant on one location from one provider is a huge cyber security risk that we think enterprises need to take more seriously,” said O’Neill. “While Amazon have a keen focus on performance, as we have seen multiple times in the last few years, a single point of failure for critical services like Auth0 and others means all services and APIs can do down at once.”

“Edge internet performance has changed a lot since our first report,” said Paul Cray, Head of Machine Learning and Artificial Intelligence. “It brings up another risk for naïve operations teams. Just looking at the connection time to get a response back from the internet could be hugely misleading. A business needs to look at the total transaction times, or else some terrible hosting choices could be made.”

The first report from APImetrics in 2017 showed a 10x difference between South America and Europe in median TCP Connect Time. Today, South America is now 40% quicker than Europe and over 3x faster than North America but it remains about 50% slower for the full transaction than the same call from North America.

“There is a belief that the cloud is fairly flat and homogenous,” O’Neill said. “It is not. Where an API is hosted, and what it is hosted on, can have a huge impact on operational performance and the performance of customers using the APIs. We frequently see that unhappy customers are unhappy because of choices they have made in their architecture, and it is hard for API providers to provide satisfaction to them when the problem lies external to the API itself.”

“One of the most important developments since 2020,’ said Cray, “has been the improvements in Connect Time in all regions, but especially traditionally ‘peripheral’ regions such as Oceania and especially South America, where TCP Connect Time decreased from 113 ms in 2020 to 3 ms in 2022. We ascribe this improvement to the rollout of improved infrastructure and increased local hosting of services, and anticipate that the trend will continue. It will be interesting to see if South Asia can catch up in 2023,”

The report provides a list of recommendations for optimum API performance, including the fact that 100 ms or more of latency (more than 10% of an average call time) can be down to a company’s choice of cloud. Not all clouds are the same and they change over time. API users shouldn’t be penalized tens or hundreds of milliseconds of latency because of a decision made years ago.

“It’s easy to look at a raw number like 100ms and think, ‘well that’s not bad,’ but then ignore the implications,” O’Neill said. “If there is a workflow with ten critical steps and each takes 100 ms longer than expected, that’s a second per transaction. If that’s across 1,000 transactions an hour, that’s now six hours of additional compute resources a day handling this. Cloud resources cost money. Now imagine 10,000 or 10 million!”

APImetrics is focused on making the data available to wider audiences and giving everyone a way to understand their cloud exposure and the risks they might face with their architectures. The API Directory, which is launching this month on the APImetrics website, will be the first step at making the data publicly available.

“We can’t pretend the business landscape hasn’t changed,” O’Neill said. “APImetrics is here to give people real understanding of how to optimize their cost base and understand that they are getting value for money.”

Marketing Technology News: MarTech Interview with Denise Vu Broady, CMO at Collibra