Blog Post

Cloud Cannibalism: Is PaaS Killing SaaS?

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

Let’s be honest: The cloud is a killer. Other than the microprocessor, I think we would be hard-pressed to find another technological innovation that has so effectively killed off its predecessors. For example — try to buy a piece of software packaged in an actual physical box. Sure, a handful of consumer-oriented PC applications are still sold in retail stores, but with enterprise applications — good luck. Even Microsoft (s MSFT) is pushing customers, both business and consumer, to the cloud for their products.

But is the cloud upping the ante, from tech killer to cannibal?

One of the earliest innovations brought about by cloud computing’s rise was Software-as-a-Service (SaaS). From to Google (s GOOG) Apps, SaaS has been the savior of small businesses and cash-strapped enterprises…but this notion is changing. XaaS has taken the spotlight from SaaS – everything from Platform-as-a-Service (PaaS) to Infrastructure-as-a-Service (IaaS) is eating away at SaaS dominance, leading many businesses to reconsider whether investing in SaaS is the right choice.

Don’t get me wrong, SaaS certainly has its upside. It’s low-cost and effectively turnkey, requiring only an Internet connection for the solution to be up and running. This great strength of SaaS, however, is also its biggest weakness — the product doesn’t actually belong to the end users. A typical SaaS solution is also relatively inflexible, requiring lengthy and expensive customization — great for plugging an application gap, not so great for building a business around. Just a few years ago, the downsides to SaaS were rarely discussed, and the alternative — custom application development — was seen as absurdly expensive and time-consuming, putting even the customization costs of SaaS to shame.

But that mindset is changing as other XaaSes, particularly PaaS, gain traction in the enterprise. PaaS essentially gives IT teams access to world-class development environments and infrastructure for what amounts to pennies, turning the old argument against custom application development on its head. Specifically, rapid application development (RAD) PaaS platforms are turning up the heat on inflexible SaaS, by speeding up custom development and slowly pushing SaaS out the door.

Essentially, RAD PaaS is comprised of two pieces: an IDE coupled with all the tools to manage the entire application delivery cycle (PaaS), a model-driven development language (RAD).  For the former, the IDE must seamlessly integrate with back-end deployment and management tools, allowing IT teams to streamline the entire application development, delivery and maintenance processes.  This exposes a whole host of functions, from configuration management and application deployment to monitoring and access control, which are delivered as services.

For the RAD component, a model-driven development language is layered on top of the IDE, dramatically simplifying the development process and allowing the development team to focus on the business, not the technology, problem.  The result is a streamlined development process that empowers not only IT, but also business stakeholders, to build applications incredibly fast, delivering working functionality in hours rather than weeks or months.

But why is RAD PaaS eating away at SaaS?

There are really three issues at hand for why PaaS, and RAD PaaS in particular, are providing a compelling alternative to SaaS.

  • Speed. Thanks to a set of seamlessly integrated services addressing the entire application development lifecycle, RAD PaaS streamlines custom development. This dramatically improves the pace of all aspects of the process, from access to the development, test and production environments to actually building and delivering the working application.
  • Cost. RAD PaaS cuts costs in several ways.  First, overall team size tends to be smaller, as the environment automates tasks typically completed by specialists, improving the efficiency of the full lifecycle of application delivery.  Next, economies of scale come into play when using cloud infrastructure, saving precious dollars.  Finally, custom applications provide a differentiator to the business at-large, providing a distinct value that’s difficult to calculate.
  • Flexibility. Through RAD PaaS, businesses gain the ability to deliver application functionality to accurately meet business needs.  This means no more suffering from misfit packages or struggling to extend existing functionality (a costly effort to undertake and maintain). Instead, custom applications now evolve with the business and stay almost ‘evergreen,’ resulting in applications that age very little and are inexpensive to change.

PaaS cannibalizing SaaS should not be looked at as a bad thing for the cloud – SaaS solutions will remain in play for many commodity applications across both small-scale operations and for larger enterprises, but their promise of speed to delivery and ability to scale is no longer a key differentiator.

The future of the cloud, and its next round of innovation, lies with RAD PaaS providing the ability to build applications, anywhere, at any time as fast as you can procure and configure a SaaS package.

And if the cloud’s future is hungry, shouldn’t we feed it?

Mike Jones is a 20 year veteran of the IT industry and currently serves as vice president of marketing for OutSystems, a provider of application development and platform-as-a-service solutions.

67 Responses to “Cloud Cannibalism: Is PaaS Killing SaaS?”

  1. Hmmmm…when someone from a PaaS company talks about the death of SaaS at the hands of PaaS, one has to question the objectivity of the one making such claims, no? As it is today Salesforce is eating the PaaS vendors lunch, so any claims to the death of SaaS seem premature and self serving.

  2. Godard Abel

    From my experience running a business using only SaaS tools with no custom business apps, the true power of SaaS and “simple” PaaS for most businesses is that they don’t need developers or development tools at all to support their own business processes. SaaS applications such as CRM and HCM provide both best practice business workflow and business content (e.g. text samples for annual reviews), so that companies don’t have to reinvent the wheel in their IT applications and can focus on driving whatever their true core competence and differentiator is (likely in their own product development, operations, and sales/marketing rather than in IT). Starting to code from scratch even on a PaaS platform seems like a slow, expensive way to go for most businesses, when a business can instead start with best practice process and content already embedded in a SaaS application, and then just configure them from there to the extent they truly need to customize. Usually companies are better off migrating to more standard business processes than trying to re-code their legacy processes. Most businesses can likely work with just SaaS tools with configurable PaaS extensions (e.g. building custom objects and work lows on that can be managed by business analysts with no need for in-house IT or development. Thus supporting custom development with PaaS tools does not really seem like a quantum step forward for most businesses, that would likely be better off by going to a zero development and zero in-house IT model which gives them greater cost leverage by eliminating all development costs and freeing them to focus on whatever truly drives differentiation in their business.

  3. Increasingly, more SaaS solutions are being architected on PaaS or IaaS foundations. I’ve witnessed and know of a number of such solutions that were built with ridiculously small teams and short periods of time. So, when your SaaS solution isn’t quite right for your business or the costs are adding up, with PaaS/SaaS and the right talent, you may well only be 1 developer/6 mths away from your own customized, future-free solution.

  4. Robb Gilmore

    I don’t believe the tools are the roadblock in most cases. Unclear requirements, and bad development process, are the major obstacles. PaaS won’t help with either of those things.

  5. Chris Riley, ECMp, IOAp, CAPTUREp, v-TSP

    I work for CloudShare which is a PaaS provider, and can tell you that the applications for SaaS and PaaS are significantly different that there are business cases for both. They may overlap for companies on the fence of moving to the cloud and liking the more control with PaaS but in general very different. This article is misleading for sure.

  6. Mike,

    Sorry, but I just don’t understand what you are getting at (unless it is a sales pitch – which is fine). The concept of “cloud” was first popularised by John Gage of Sun in a very obscure but pertinent statement – the network is the computer. The recent “boom” in cloud offerings is an extension of the success of SaaS vendors (pls note the “software” in SaaS). The variations of PaaS and IaaS are just extensions of the generic Software as a Service concept. So, how does question of SaaS being “replaced” by other XaaS even arise is beyond me. The early software vendors used to provide the infrastructure (servers etc) and functional software as one offering. Now, vendors are differentiating by offering specific services that users can combine to offer their own specialised or customised offerings – so where is the threat? I can understand if you say that is under threat due to the wider variety of options now being made available by both software vendor/ISVs as well as hardware vendors but stating that SaaS is being eaten up by Xaas is really missing the point.

    • Mike Jones

      I think there is a great debate in your statement that the current cloud offerings are an extension of the success of SaaS vendors. Putting that aside, I believe that the key in all of the viable cloud offerings is the word Service. That being said I think the notion of cannibalism and PaaS killing SaaS worked to get attention but ultimately clouded (no pun intended) my real intention with this article. My simple point is that the wave of emerging and rapidly maturing PaaS offerings are providing a custom development alternative which shifts the playing field. Thus making custom development more of an alternative to SaaS than it has been in the past. In a recent discussion with a prominent Gartner analyst I captured the following: The Business purchased SaaS because they did not want to deal with IT. IT purchased IaaS because they did not want to deal with procurement. We did not come up with a statement for PaaS – I guess time will tell. Perhaps it will be something like – The Business and IT purchased PaaS because they needed a specific solution fast.

  7. Joanne Moretti

    It’s the same as before: buy versus build, that is the real decision. What does your business need, what is your business risk level, how differentiated do you want to be, etc Those are the questions, not Paas versus Saas.

  8. If a SaaS solution gets me 80% of the way to my ultimate solution, why would I throw that away to rebuild some ultimately-customized solution when I could just take it that last 20% of the way in-house and save all the effort that went into building the SaaS solution in the first place?

    I think the article dramatically underestimates resource costs in planning and executing IT and software projects. If the project is so easy that it can be up and running in a day, I have a very hard time believing it will capture all the business requirements for all but the most trivial of applications. Even in the age of RAD and Agile, planning and testing are essential to well-executed, useful IT projects. Furthermore, abstractions are leaky:

    When you can’t quite get the business logic right because you’re boxed into what the RAD framework provides and you hack a workaround together that doesn’t quite work, how do you dive into the stack or dig into the problem? It may get you started, but I don’t think it’s tenable for a long-term solution. I think the solutions described will be overwhelmed by the complexity of the business’s requirements, and require the company to ultimately build their own SaaS solution in-house.

  9. Hello, I tend to disagree. I do believe that both SaaS and Pass will survive. The ones in need for PaaS like flexibility would have never considered a SaaS but rather delpoyed on premise. SaaS still makes a lot of sense for highly standardized (not say commodity software) where there is no need for customization any more. Some will still build out their own special thing but that was the same iwth all technology changes.
    On top of it why should I have my specialzed folks build something on PaaS which is available as SaaS when they can build the customized business app I need. It is about focus and PasS is great but it is not about cannibalizing.

  10. Interesting post. I would like to disagree with the fact that PaaS is killing SaaS but it may broaden the scope of the market. PaaS is an alternative for businesses which do not find an exact match with the existing SaaS applications, though lot of SaaS apps can be customized. PaaS on the other hand helps to create custom SaaS applications. See as an example of RAD PaaS.

  11. I Am OnDemand

    The title of the article made me read it though while reading the comments it seems that the writer intention is marketing. This confusing. I will relate to the question b saying that there are several PaaS vendors that are going towards offering apps/plug-in models for the DevOps team. This enables the enterprise to create semi-custom made applications. My thoughts on that matter are still evolving though for now I tend to agree with Bert’s comment above. Ofir

  12. Paas and Saas are two different solutions for two different problems. If the application you want in the cloud is commodity and doesn’t change often (a mail server like hosted Exchnage for instance) then SaaS is clearly a better option. If its less commodity, more bespoke and/or you want to update/release it more often then PaaS can be a better option. PaaS isn’t killing SaaS were just seeing a bit of settling as people ome to understand the distinctions more clearly and adjust their models.

  13. I personally think and believe that paas could pose an increasing competition to saas . cannibalising may be a relative term.

    A lot needs to be done at the paas side from today’s incipient stages. its good to see rapidly evolving paas such as cloudfoundry,cloudbees,appfog,redhat flex etc…

  14. Allen Kent

    We ( have provided server-based medical scheduling software, extensively configurable and easily customized since May, 1998. Promising new customer acquisitions grow with every new acronym: ASP, SaaS, the Cloud. Our policy to restrict feature development to work customers are willing to pay for, that is, to read real needs, means our offering improves with each new customer. We compete with the medical community’s tendency to “hire more help to get out the schedule!”. We think our sever-based software (SBS?) is gaining on ’em. -Managing Partner

  15. The cost of the hardware platform was never the only argument against custom application development. People costs are usually the biggest expense by far. Time to delivery, quality of the delivered application and on-going maintenance add to the costs and uncertainties associated with custom development.

  16. Scott Johnson

    Agree with ShillShocked, Adam and others. Statements are being made as if they are facts, yet no mention is made of where the data came from. Truth is, the existing data from Gartner, IDC and Forrester all contradict the trends being forecast here. I thought GigaOm was better than this.

    • Mike Jones

      I did intend to create a bit of controversy and discussion with this piece. While I only have access to Gartner and Forrester data my research shows that PaaS, while a very small part of the overall cloud market today, is slated to grow at a much faster rate than IaaS and SaaS over the next few years.

  17. Sure we now have a platform for developing apps, but it was never the platform that cost money and time, it was the people doing the development. Unless we are talking about putting developers “in the cloud” (it was once called contracting), then nothing much has changed.

    • Mike Jones

      Simon , I think you are absolutely correct – the cost of development has always been a people issue. The opportunity before us is to change the way people build software. Don’t wait for operations to procure the development, test and production servers. Get started while the business has the problem. Don’t wait for the business to submit their full requirements. Get started building something based on the business ultimate goals and vision. Don’t be afraid to change direction if it furthers addressing the ultimate goal. Get a development platform that lets you change fast with minimal risk.

      This new wave of RAD PaaS solutions provide the tooling needed to address these goals. Thus reducing the people issues associated with development.

      • Alex Charles

        The answer to your “start now” approach is more about development methodology and culture than a technical limitation.Good Agile houses will have product out of the door in weeks whatever the technical environment.

  18. Michael Bordash

    We offer a PaaS in the mobile space. If we were to create a pure SaaS offering we’d be dead in the water and we’d not be able to keep up with the daily trends happening in mobile application development. Our PaaS allows us to view how developers are taking our core IDE and platform to levels we haven’t begun to think about yet. In this way, we also get pretty good idea what our product roadmap should look like… common trends within our development community will be adopted into our core. Great article.

  19. So you bring up and Google Apps as examples of Saas. Do you really think writing a CRM system or a productivity suite from scratch on a Paas platform is going to be cheaper, or have any other benefits??

    • Mike Jones

      Great question – the answer lies in the business need. I like to refer to Geoffrey Moore’s core / context model. If the application is truly serving a non competitive ‘context’ business process then SaaS is perfect. The issue is that in many cases the business need is not completely met by the SaaS package thus some form of customization is needed to the SaaS offer or the business must adjust its process to fit the package.

      I have seen the mis-alignment of the business need to the SaaS package so great that custom development is the only answer. A recent example was a customer who had a very specific need for supporting customer relationships – sounds like CRM. When the dust settled the SaaS CRM was completely replaced as the needed functionality was so small from the SaaS it only took 5 days of additional development to replace it using the RAD PaaS development solution. This customer turned off their monthly SaaS subscription and had an immediate ROI including both the cost of the RAD PaaS solution and the development expense.

      • Mahesh Paolini-Subramanya

        Still think this is a bit of a strawman. You can call it core/context, or service/platform, but at the end of the day you are still talking about two fairly distinct things that happen to intersect in some areas. Your example could be turned around by quoting somebody else who built their own ‘CRM’ solution using a PaaS, when they could have used SF….

    • Mike Jones

      Hi Paul, the PaaS market is rapidly evolving and new entries are showing up almost daily. Some of the well known vendors include, Google’s App Engine, Microsoft’s Azure. Then there are some emerging players like Longjump, Wavemaker, Mendix and of course (under full disclosure – I work for OutSystems) OutSystem’s Agile Platform. Note, in my opinion, RAD PaaS is striving to go beyond what many of the current PaaS vendors offer by providing a model driven approach to application development that insulates the development team from the underlying technology stack, provides all the tools to not only build the application but to deploy it, monitor it, scale it and change it all from a simple to use console that surfaces the needed services. Thus a true set of services tightly integrated into the PaaS offering.

  20. Tom Halpin

    The term ‘cloud’ is mis-used in many cases, for sure – however in the correct context, cloud does not relate to VPS system individually. Its the combination of multiple virtualised environments that provides a set of computing resources, distinct and independent (not tied to) of physical servers. Renting a VPS is not a server in the cloud per-se – however renting a VPS instance which has seamless mobility/failover on a virtualised platform could be regarded as a server in the cloud. IaaS is an extension of that concept (on-demand provisioning, wholesale rental and utility-model billing)

    • computing resources not tied to physical servers? So you can get more computing resources from the cloud than you can from the physical servers that made up the cloud? jk. What kind of “mobility” are you talking about?

      My point is that the resources have always been there. They recently became mainstream because businesses found new ways to make loads of money.

      • Geert Bellemans

        What Tom Halphin is saying (I guess) is that your virtual server is not tied to a physical server so when you run out of resources, there’s no admin that needs to insert extra RAM (or some other resource) in some physical server, it’s just assigned to you dynamically (in a way that’s transparent to the end user). Thus, the other way around: when you’re not intensively using your cloud server (low load) the remaining resources can be assigned to another end user, this reduces the TCO of the physical servers.
        So when using IaaS, your server instance is ‘somewhere’ on ‘some server’, and it can be ‘jumping around’ across servers without the users needing to know about it –> it’s mobile.

        This way you can get a lot more out of your physical servers because much less resources are wasted, it’s true that this is a new lucrative business model but in the end it’s also more cost effective.

  21. Shock horror. VP of company selling something I’ve never heard of before (RAD PaaS), suggests this new thing is the future.
    Or in this case that is, so long as Windows/IIS/SQL Server or Oracle is suitable as complete solution. One of the big strength of SaaS is that infrastructure & technologies can change over time without any changes to end user. Seems like this has some plus, but adds a big vendor + technology lock in.

    • As I was reading the article I thought, “I wonder what horse this guy has in the race?” Then I get to the punchline. Not only is he not a technician (VP of Marketing), but it’s for a PaaS company. Yeah, totally objective article…

    • Filip Vuksanović

      I read about this:
      “dramatically simplifying the development process and allowing the development team to focus on the business, not the technology, problem. The result …., but also business stakeholders, to build applications incredibly fast, delivering working functionality in hours rather than weeks or months.”

      for last 15 years. Every silver bullet wannabe RAD / Framework has that line in its marketing pamphlet. Still,
      what is really changed in this area since SaaS introduction?

  22. From what I understand, the cloud has been around for a while. It’s just a marketing term for a bunch of vps (or for large companies a bunch of dedicated servers) put together. Am I wrong? If I’m not, how is the could a “technological innovation?”

    • it’s not just a marketing term. it’s all about the ease of provisioning machines. think of the difference between putting in a purchase order for another server versus right click -> launch new instance. it’s outsourced virtualization.