LinkedIn Twitter
Director, OpenShift Strategy at Red Hat. Founder of Rishidot Research, a research community focused on services world. His focus is on Platform Services, Infrastructure and the role of Open Source in the services era. Krish has been writing @ CloudAve from its inception and had also been part of GigaOm Pro Analyst Group. The opinions expressed here are his own and are neither representative of his employer, Red Hat, nor CloudAve, nor its sponsors.

11 responses to “Next Iteration Of PaaS: Will Amazon Join That Race?”

  1. JP Morgenthal

    Krish,

    I agree with the direction for your vision for PaaS 2.0. Clearly, the whole value proposition is the availability of consumable, metered services that are available to the PaaS container and shared across all instances. Data management is one of these services as is Security-as-a-Service (SecaaS), Messaging-as-a-Service, Collaboration-as-a-Service, etc. the goal is to make it as simple as possible for developers to build out new applications using basic semantics and let the power of the PaaS provide the speed, scalability and elasticity.

    -JPM
    http://about.me/jpmorgenthal

  2. Steve Ardire

    Hi Krish,

    Good post !

    Yes PaaS is evolving to next gen intelligent applications and PaaS 2.0 has “Data as a Service” component tightly integrated within. Then I see ‘Linked Data as a Service’ http://bit.ly/tPNeHq coming into play as a follow on.

    One question i.e. when you say only handful of cloud providers ( Amazon and ? ) will be in the game I assume you’re talking about public clouds but not hybrid or private clouds

    – Steve

  3. Sinclair Schuller

    Krish, good post. I agree with your overall position, although I’ve thought about it a bit differently. Data and the ability to work with data will undoubtedly become the most important part of PaaS 2.0, but I don’t think an analytics engine is the hub in that sort of system. Instead, I believe that *good* PaaS offerings will offer trivial support for advanced patterns and computation models. This will be the preferred approach, which can equip developers to trivially build powerful analytics engines that are highly specialized and tuned for their data sets. Essentially, we’ll see a transition from multi-threaded apps as being the thing that “cool developers” do to multi-agent systems (MAS – something like what is described here, but I see a slightly different outcome to help broker the transition from old paradigms – http://en.wikipedia.org/wiki/Multi-agent_system). The PaaS offerings will make building MASs a simple matter, allowing developers to focus on the logic executed by these MAS engines.

    The reason this distinction is important is that MAS is a general purpose construct that even in non-data intensive scenarios can be a powerful execution backdrop, and is a bit of a DSL for the analytics problem.

    Not to beat a dead horse, but this is why I think that a deep runtime model for PaaS is appropriate rather than the non-runtime “Look at how many languages I support” approach; a runtime model is required to get this level of richness in the execution model, and the level of instrumentation necessary require focus on depth by PaaS providers.

    Just my 2 cents.

  4. Derick Townsend

    I like the innovative thinking here, and believe that data-as-a-service will become a more sophisticated component of PaaS. It wasn’t clear to me if your PaaS 2.0 definition was encompassing the future of public PaaS offerings, private PaaS offerings, or both. If the focus was primarily on public PaaS, I have a couple of concerns about broad scale adoption of certain use cases you mention:

    1) To realize the vision of self-optimizing applications with advanced data analytics, you’re talking about a lot of data over time. If you are dealing with a federated model of public clouds as you mention, you’re talking about potentially very costly bandwidth to move that data – perhaps cost prohibited. If everything was under one cloud, you could minimize bandwidth consumption, but you’ve vision leans towards federated clouds, which BTW I totally agree with.

    2) The underlying raw data (for example, to support customer analytics) could represent a very strategic asset to an enterprise such that they may not want to move it off-premises… either due to regulatory constraints, fear of loss or theft, vendor lock-in, or other concerns. Corporate data security and governance concerns can be big hurdles to clear.

    For these reasons, I wouldn’t minimize the importance of private PaaS 2.0 offerings with powerful analytics, etc. In fact, we are seeing growing interest in more advanced private PaaS in large enterprise accounts – http://bit.ly/qNHClu. My 2 cents and best of luck with the fine-tuning.

    DT

  5. Remember Next Gen PaaS and AWS? Here Is The Second Piece To The Puzzle

    [...] Krishnan Subramanian on February 21, 2012 When Amazon announced the release of DynamoDB, I argued that it was their first step towards joining the PaaS game in its next iteration. I am completely [...]

  6. Next Iteration Of PaaS: Microsoft Game Plan

    [...] on March 5, 2012 In January, I proposed a simple model for the next iteration of PaaS, called Intelligent Platforms, which is centered around data. As we move into a world dominated by Big Data with mobile and [...]