More than two years ago I wrote the blog post "PaaS Comes Of Age," where I posited we were entering a platform-as-a-service era -- one in which platform services would become the critical cloud capability for enterprises. A lot has happened since I wrote that post, and now we are seeing this "coming of age" story continue into maturity.
For the uninitiated, PaaS can be thought of as a delivery model for functionality you traditionally associate with middleware. Not so coincidently, it also sits in the "middle" of the cloud stack, between infrastructure-as-a-service and software-as-a-service. PaaS can deliver any combination of application "development" and "runtime" services from the cloud, in some type of measured model (subscription, pay-as-you-go, etc).
Today there are quite a variety of PaaS services, so many that it's hard to keep them straight. Many companies have converged on this space, coming up from the IaaS layer (Amazon's Relational Database Service or Elastic Beanstalk), down from the SaaS layer (Salesforce's Force.com), or even as a new corporate initiative (Google AppEngine or Microsoft Azure). In addition, there are numerous "pure play" PaaS offerings (like EngineYard) and a whole slew of startups.
One aspect to watch carefully in the PaaS space is the role of open standards and open source. Unlike IaaS, where Amazon has become a bit of a de facto standard, PaaS is likely to be more de jure -- built on a set of open standards. IBM just announced BlueMix and a $1 billion commitment to PaaS capabilities. Key to this announcement was the support of Cloud Foundry, an open-source platform layer for the cloud. Started by VMware and spun off as part of Pivotal, Cloud Foundry already has around 40 corporate contributors. Not to be outdone, RedHat's OpenShift initiative is also based on open source and also has a stable of active contributors.
[Want more on Pivotal's latest moves? Read Pivotal Brings In-Memory Analysis To Hadoop.]
It will be interesting to see what the future holds for PaaS. If it tracks with similar technology shifts in IT history, here are some things to assume.
- Companies need to become comfortable with the concept before they standardize on a PaaS. We are still in the tire-kicking stage in most enterprises, which want to see the value in PaaS before starting any serious shift off traditional middleware. This would imply a year or two before real PaaS growth. But, much like the shift from client/server applications to Internet-based applications, when the shift starts it will ramp quickly.
- Open systems will coexist with proprietary. Much like Java and .Net, there will be followers of each.
- Developers will decide. Think early days of Microsoft or today's Android. The PaaS vendors that are most successful in courting and keeping software developers will rise to the top. Don't be surprised if you start seeing some "PaaS-justifying applications" -- compelling SaaS offerings that run only on a particular vendor's PaaS. (For the gamers out there, an analogy is Titanfall for Xbox One.)
- Consolidation will occur, but not as much in the PaaS market as in IaaS. Some believe that IaaS vendors could consolidate down to as few as six major ones.
- If you look at today's analogue -- the server and storage business -- this seems about right. But PaaS is a much broader set of capabilities, and one in which ecosystems can develop around standards (much like the Java community today), but still have room for a larger field of competitors.
I will check back in two more years and see if these predictions prove to be accurate.
Private clouds are moving rapidly from concept to production. But some fears about expertise and integration still linger. Also in the Private Clouds Step Up issue of InformationWeek: The public cloud and the steam engine have more in common than you might think. (Free registration required.)