PaaS simply isn’t used as much by enterprises as initially predicted. Although PaaS offerings vary greatly, most provide facilities for application design, deployment, testing, and self-provisioned hosting. It’s best described as a turnkey development and deployment solution.
Why hasn’t PaaS taken off? I see several reasons.
- Strong tools from IaaS providers, such as Amazon Web Services and Microsoft Azure, seem to be getting enterprises’ main attention and investment. Even when broad cloud platforms have PaaS as part of their tool set, most developers opt for core IaaS, such as storage and compute platforms.
- Developers do not like to be confined in a sandbox. Many PaaS providers impose restrictions—such as tools, databases, and programming languages—that developers typically don’t prefer.
- IaaS seems to be a better fit than PaaS for devops organizations because it provides the operational platforms (what the "ops" part in devops manages) that developers will ultimately use: those in the IaaS platform, as well as their in-premises environments, which they can replicate in an IaaS cloud.
Back in 2008, when the cloud was new, the federal government and industry partners defined what became PaaS to address a real need of that era. That need still exists, but today PaaS has become the most ill-defined area of cloud computing as the industry broke it apart into narrow slices.
As a result, approaches, features, and definitions vary widely, with many PaaS providers offering a specific focus. Some, like Salesforce.com’s Heroku, focus on specific programming languages, such as Ruby, Node.js, Python, or Java. Some, like Oracle’s Cloud Platform, focus on tight integration with major databases. Some, like Cloud Foundry, focus on a specific standard.
To make matters worse, most take their own propriety, closed approach to PaaS, even if they say they are following a standard.
Still, I don’t think PaaS will die from these self-inflicted wounds. Instead, I believe it will be absorbed into IaaS platforms even more. Global 2000 enterprises will lose interest in the smaller PaaS providers, considering them to be risky platforms to base production work on for the long term. The larger cloud providers will do what larger providers do: Make related functions like PaaS systemic to their overarching (IaaS) platform. You can bet that AWS, Google, and Microsoft will all do this.
PaaS will live on as a component of the cloud for building applications, but it’s essentially dead as a standalone service. Farewell, PaaS; we hardly knew ye.