The other day I came across this awesome post from Peter Sonsini from NEA. I’m actually loathe to critique it – Peter has gone to lengths to present a coherent thought leadership piece on the future of PaaS and PaaS providers – there are so few people who truly understand this space that it seems a little churlish to try and negate the thoughts of one of those people. But there ya go, the joys of being a commentator huh?

Basically Sonsini talked about the massive opportunity he sees for PaaS providers to be the curators of services for their customers. To put it bluntly, Sonsini predicts a future where PaaS vendors will differentiate the rubbish add-on services from the solid ones, and deliver up the solid ones to their customers in some kind of marketplace. So far so good right? Well maybe, but it misses the point around cloud services.

In the infrastructure space we’ve reached a fairly common acceptance (well at least those not wedded to a “single vendor for everything” strategy) that the future will be heterogeneous with organizations leveraging the services of different cloud stacks, different hypervisors and different fundamental approaches depending on the needs of the particular business unit or workload in question. This is summarized by my contention that if we know anything, it is that the future is heterogeneous.

So let’s look at that for a second. Core infrastructure, delivered by seasoned IT practitioners who have the time and skills to wrangle solutions to fit a particular type of plumbing, will be heterogeneous in nature. Well if we accept that fact, how on earth do we expect one PaaS solution to meet the needs of every different use case and business unit within the organization? The fact is, the further one goes up the tack, the less likely it is that one solution will meet all the organizational needs. By extension, and logically following that thesis, clearly PaaS, like IaaS will be heterogeneous in nature.

If that is in fact the case, is it really either appropriate or effective for the PaaS vendor to be the gatekeeper to the services the applications on the PaaS uses? Or the management of the PaaS workloads? It seems to me not. In the same way that an organization’s infrastructural management is ideally managed by a third party service such as enStratus, so to are an organization’s PaaS assets, the services that plug into those PaaS’ and the different instances of all the different PaaS’ that the origination uses most ideally delivered by a third party provider.

It’s something I’m actually quite passionate about – not using management tools that force business units to adopt a particular solution that may or may not really meet their needs. It makes no sense and actually hampers the ability for cloud to deliver the benefits it promises.

Which brings me around, a little reluctantly, to Appsecute. I’m always loathe to write posts that seem to be self-serving to companies I invest in. But on the other hand I invest in companies precisely because I believe they solve a real problem. Appsecute clearly fits into that bucket. I have a couple of very strong beliefs as articulated previously:

  • PaaS is the medium to long-term future of cloud services
  • That future will be wildly heterogeneous

Put those two things together and what do you have? A requirement for a solution that gives an IT organization a broad management layer across multiple PaaS’ and multiple services that can be used with those PaaS’. Sonsini was so right in so many ways, but so wrong in so many others. Yes PaaS is a fantastic deliverer of different functional services – as it should be. But individual PaaS providers should not be the arbiters and managers of those PaaS services, rather a broad management layer is required to fulfill this need.

Ben Kepes

Ben Kepes is a technology evangelist, an investor, a commentator and a business adviser. Ben covers the convergence of technology, mobile, ubiquity and agility, all enabled by the Cloud. His areas of interest extend to enterprise software, software integration, financial/accounting software, platforms and infrastructure as well as articulating technology simply for everyday users.

  • Ben, good post.

    Now, I am not sure I entirely agree with you 🙂

    If you look at the middleware space, which is the closest we can look at when it comes to PaaS, the market has gone through tremendous consolidation over the last decade. Take Java, a market that I know a bit, we’ve drifted from about 30 vendors in the early 2000 to less than a handful today (IBM, ORCL, RHT and VMW). Consolidation will happen on that strategic layer, and ecosystems (who are economically not interested in integrating with a dozen vendors, but typically 2-3 at most) will be accelerating that feedback loop.

    Now, you are right, PaaS providers will never become the best at providing everything developers need. Much like ISVs are providing best-of-breed certified solutions around/on top of middleware solutions, the same is and will be true in the PaaS space: best of breed “SaaS” providers will complete the base PaaS offering.

    Consequently, if by “curated”, it means that those 3rd party integrations will “just work” (i.e. deep integration, well tested, documented, etc.), then I don’t think developers will see a problem with that, au contraire. On there other hand, if “curated” means “protecting a PaaS backyard”, providing limited choice and limited ability to extend a PaaS in a self-service fashion, then I don’t see this fly with developers.

    So I don’t think that the freedom to extend a PaaS without constraint and the notion of “curated environments” are at odd.


    Sacha Labourey
    CloudBees, Inc.

  • I agree that PaaS is the medium to long-term future of cloud services and that it will be heterogeneous because of the simple fact that organizations will leverage different types of cloud solutions to best suit their needs. This means that ultimately, the PaaS vendors have to offer different types of cloud services to stay in business.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.