11436 SSO

API Platforms: Buy, or Roll Your Own?

Michael Leppitsch
Apr 30, 2014

What’s the cost of construction and maintenance of an API infrastructure?

An architect at a major enterprise recently explored the idea of creating a spreadsheet to compare the cost of building an API management platform to the price of licensing Apigee’s product. This type of static cost comparison is all but impossible in a world where the technical tools and platforms evolve at the blazing speed of internet innovation itself.  

The construction is only a small part of the total cost of ownership, but even that part is like doing algebra in a calculus world. Follow along with this narrative of estimations as we look at the dynamics at work.

Playing catch-up

To start, Apigee is an end-to-end, full-featured, integrated platform that benefits from multiple iterations of development, and the learnings distilled from hundreds of enterprise deployments. The latest generation alone is the result of hundreds of developer years, creating a highly refined product that’s available to drive business value in today’s competitive digital world, right now.

Assuming one finds an amazing team of rock stars that can reach functional parity at half the effort, leveraging all the open-source software Apigee has made available to the world, and needing no generational restarts, this would leave over a hundred developer-years to catch up.

Of course, during the two to three years it takes an enterprise to do this, Apigee will have doubled the level of automation, functionality, analytics, tooling, integration, and add-on libraries that’ll be required to compete in that future world, three years from now. All of this will be available with a simple license purchase. Essentially, the DIY team is likely to stay hundreds of developer-years behind.

A battle-hardened platform

Meanwhile, the Apigee platform becomes increasingly battle-hardened by billions of API calls per day. Countless efficiencies and tuning changes are implemented and tested in real-world scenarios, to the benefit of hundreds of Fortune 1000 companies.

Assuming you could gather up a team of 50 to 100 developer rockstars (and you probably could, with enough money) who are up to the challenge of building and maintaining an in-house, feature-rich API management platform,  these same rockstars would never attempt this at their own startup.

Instead, they would attack a totally new problem in an area that doesn’t already have market leaders like Apigee, which is now moving on to disrupt adjacencies like predictive analytics. Me-too technology business models simply don’t make much sense.

Operational challenges in a DIY environment

In addition to building and operationalizing a DIY platform, one would still need a separate API design team to implement the actual APIs, configurations, API policies, and back-end upgrades to create the actual business value to project into the world. Instead of doing this work on the world's leading API platform, this team would be doing it with a proprietary and internally built and maintained toolset, one that would require proprietary skills that aren’t transferable. History shows that such teams are very difficult to recruit for.  

Because of the requirements demanded by operating in the digital ecosystem and the breakneck pace of innovation, a side-by-side cost comparison isn’t really possible. But by juxtaposing the multiple and significant waves of time and money required to build a proprietary API platform against investing in an established, tested, market-leading product, the answer to the build versus buy question becomes quite clear.

Image: 3SugarCubes/Flickr


Scaling Microservices