API Science Tool Marks Maturity of API Provider and Consumer Market

API Science is a cloud-based API monitoring service that embodies a new generation of API tools that support API runtime production operations. The service lets API developer-consumers track when an API call does not return the needed data or functionality to power their apps. Consumers can monitor performance speed, availability, and data quality and set regular timing of monitoring from global locations. For API providers, the monitoring service is enabling an outside-in look at what really happens with their API when it is in production at a global scale.

API Science

“Companies are now starting to realize that you need API monitoring from the outside-in so you can see the real health of your APIs,” says John Musser, founder of API Science (and founder of ProgrammableWeb). “It’s not until you run into some issues with your APIs that you realize you need a better process to know your APIs are up. Even if you are not providing public APIs — if you only have internal APIs — you still only discover this out of necessity.”

Anatomy of API Science

API Science allows users to monitor a range of their own or external APIs by adding the API endpoint and any authentication parameters. Users can also import APIs using RAML, Swagger, HAR or Postman importers.

API Science dashboard

API Providers Monitoring Private, Partner and Public APIs

Musser says he first got the idea for the monitoring service when he was leading ProgrammableWeb, where API providers being able to see what their developer communities were experiencing became a growing concern, “but the market has only matured enough in the past two to three years that there are now enough customers wanting to buy this sort of monitoring service,” Musser explains.

Like many tools emerging as ancillary services to API management, Musser is seeing a greater need amongst customers looking for ways to monitor private and partner APIs over public-facing products. “But if you look at three of our customers, you cover all of those API maturity scenarios,” says Musser.

He points to Reverb, which was recently acquired by News Inc. Their entire app infrastructure (including products and features like their mobile app and their news aggregation feed) is built on APIs: “They use API Science to do more sophisticated end-to-end-testing. They can also run JavaScript style assertions so they can look at the JSON payload and make more intelligent diagnostics on that. Previous generation tools particularly in the API space, would preempt what those APIs would return.”

Musser says this is one of the strengths of API Science that has emerged because of the growing sophistication and maturity of the API economy. API developer consumers — or businesses that dogfood their own APIs — can run multi-step performance testing, for example, starting with an authentication process, calling an API endpoint, parsing some of the data that is returned and then running that as the query parameter through another API call.

Because API Science tracks the actual performance of such multistep processes that mimic actual API usage scenarios, rather than preempting the return, the tool can provide much more powerful monitoring into where problems might occur.

Musser says other key customers like the travel expenditure platform Concur and the stock photos library GettyImagesTrack this API all have public, partner and internal use cases for their APIs that they are monitoring using API Science.

The API Science API

While API Science has its own customer dashboard and interface and enables providers and developer-consumers to track their own and external APIs through the SaaS browser interface, the API Science API also enables dev teams to integrate the monitoring tool directly into their existing systems.

“API Science has an API so you can integrate it into your deployment workflow, you can use this as part of your continuous delivery workflow and make monitoring more of a first class citizen,” says Musser.

He explains that by using a continuous delivery workflow, API providers can provision a new update to their API platform. Then they can use the API Science APITrack this API to monitor the latest changes to the provider’s APIs automatically. “Every Ops team out there has multiple dashboards, so our API can pull that monitoring data into a pre-existing dashboard or you can use it to create your own status page on your dev portal,” Musser explains.

Improving API Provider Performance

Musser describes a number of situations where API providers have been able to improve the performance and reliability of their APIs by using API Science to monitor the external experience of their API:

  • One customer was using API Science to track their API performance and discovered that after a major 2.0 release to their API, their API had actually become 30% slower on a number of key API methods. They were able to quickly identify the bottlenecks and now the 2.0 release is at least 15% faster than any prior release.
  • Another customer used API Science to do side-by-side performance comparisons of different API versions (the current and the next upcoming release). This gave them side-by-side real-world data, from multiple geographic locations, using different data sets. They ran these over a period of weeks and were able to optimize their API performance with much greater accuracy than previously possible.
  • One was able to track performance of their API from multiple locations around the world, all the way to their API gateway and then down to their database. Using API Science, they created a simple 2 line bit of JavaScript that executed at the start of the API monitor call and appended a dynamic 'correlation ID' to the end of the API call, and then they could easily match any specific inbound monitoring call to their backend monitoring systems and thus got a holistic end-to-end picture of their API performance.  
  • One customer had an API that was returning what seemed like a successful response, with an HTTP status of 200 and no error reported. But using API Science's ability to setup very specific validation rules around JSON response data, they found that it wasn't consistently returning the number of rows of JSON data that it should. It was a very subtle, hard-to-find issue that was not uncovered in conventional testing, and one that's not uncommon in real-world API scenarios.

An Ancillary Service

Musser says ancillary services like API Science that help API providers manage their APIs are becoming the norm. As the API economy matures, API providers are looking for more tooling to support their API in addition to the standard API management solution.

“Typically they run into issues for performance availability: the question really comes down to how can they most cost-effectively monitor their API health and availability?” Musser says the advantage of API Science is that it can manage a set of global monitor nodes without each API provider needing to establish their own. So far, API Science has nodes located in North America, Europe, and Asia with other locations on the way.

Musser is keen to clarify the particular role that API ancillary services like these monitoring tools are now beginning to play. They supplement API management solutions, and provide a more granular, performant level of analysis on APIs than more general SaaS monitoring tools.

Musser points to a number of specific performance issues that services like API Science need to monitor:

  • Network connectivity errors
  • Data integrity errors
  • Slow call response time
  • HTTP errors
  • Invalid JSON or XML
  • Authentication errors
  • Server availability
  • SSL errors
  • Latency spikes.

“We feel one of our strengths is how quickly you can get going. If it takes more than 60 seconds to set up your first monitor, we haven’t built it right,” Musser says.

For API Consumer-developers

While API providers can use the service to make sure their customer base has performant, reliable access to their APIs, API Science also hopes to support the needs of API developer-consumers. These users users may be consuming third party APIs to power their own apps or internal business workflows.

Musser explains:

When you sign up for an account, we have some sample monitors, but you can create a monitor on any third party API that is out there. If you use any third party service, you can design your own monitor against that, and it will live in your API dashboard. We have customers who monitor dozens of third party services.

People don’t realize how many third party APIs they are dependent on. You ask someone how many APIs you use, they will say a few, but then if you let that conversation hang a little, they will inevitably extend the list: ‘Oh that’s right, Google analytics, social media authentication… it ends being 2-4 times longer than people at first realize.

The type of service API Science provides is going to matter more and more as we continue to build-out our apps and services in the growing fabric of APIs everywhere.

Developers and API providers can sign up for a 30-day free trial at API Science. After that, pricing is tier-based, with thresholds on the number of API calls made to monitor the APIs and maximum number of team members accessing the same account. 

Mark Boyd is a ProgrammableWeb writer covering breaking news, API business strategies and models, open data, and smart cities.

Comments