APIs Connect Apps—And Slow Them Down
by October 30, 2017

Filed under: Cloud Computing, Performance Monitoring

APIs build bridges between applications. Now that companies are moving quickly to cloud-based and SaaS applications, these APIs are a lifesaver. Just one application can have any number of APIs connecting into it, creating a web of communication among important business-critical applications. This API economy has become another facet of modern infrastructure that IT teams has to manage.

An EMA report found that APIs have become common currency for IT infrastructures today, and that they present a whole new set of challenges for IT teams. Their study of companies providing APIs showed that the top three challenges were high traffic volumes, back-end security and identity management. In general, metering, chargeback and access are also problematic for API users.

Just one critical SaaS application might depend on many APIs to complete a transaction or save a lot of time for users. We took a look at Salesforce here at AppNeta, and found that just four integrations added about 4 seconds to the app load time every time. The cost of those slow load times adds up quickly.

Clearly, this web of intra-dependent APIs applications isn’t self-sustaining or hands-off. But it’s super important in a world where companies often select multiple best-of-breed apps over a solution that professes to do it all. The efficiency that comes with a smoothly running, API-connected app can really boost user productivity.

This mess of APIs is just one more complicating factor in today’s modern IT environments. Plus, the APIs themselves can vary a lot in their reliability and maturity. Don’t give up on actually managing them, though. There’s a way to do it.

How to Get Ahead of API Sprawl

When you get a call or help desk ticket submitted from a user having problems with a particular app, there are some rudimentary diagnosis methods. You might check the status page of that application, and check the status page of the ISP serving that user. Beyond that, you may have a handful of network metrics that don’t really show problems. This is all made more complicated by many of today’s single page applications, so that it’s even harder to pinpoint where something went wrong.

As in other cloud and SaaS environments, legacy monitoring tools aren’t keeping up. Most providers monitor application performance and availability from the point of view of the gateway—the front door of the application. But testing just to the gateway doesn’t take into account any of the 100+ components of that application that could be broken.

Those potential failure points can easily seem overwhelming to an IT team trying to resolve user issues. But seeing every place where performance problems occur is essential.  

Our synthetics monitoring tool specifically addresses these modern problems to go beyond the gateway. We actually track user interactions with the application, rather than playing a guessing game. Traditional synthetic scripts depend on page loads to display timing data with single-page applications. Traditional scripting can’t make the connection between latency and user actions, and don’t see much beyond a basic availability check.

Selenium scripting that cleverly folds in Milestones lets AppNeta see when a user action is complete. That means you can see timing details for any sequence of actions, even on a single-page application. So our users see exactly where a problem is occurring in a workflow, resource load or over multiple pages.