Longhaul flights are sometimes useful in that they allow one to ruminate over some of the issues that the daily deluge of data doesn’t allow. On this partiuclar flight I’ve been ruminating over a post I read recently from Yobie Benjamin. In the post he interviewed the CEO of SugarCRM, Larry Augustin. The gist of the post was the assertion by Augustin that what he calls cloud computing is going to render traditional SaaS obsolete.

First some clarification – what Augustin is really saying is that IaaS will prove so compelling that it will force SaaS vendors to either move to IaaS or fall by the way side – in other words SaaS vendors with their own infrastructure will find it increasingly impossible to compete. In Augustin’s words;

Cloud computing is obsoleting SaaS as defined by SalesForce, NetSuite and similar single-vendor solutions. The distinction between SaaS and cloud computing is an important one. Cloud computing gives companies the ability to scale out computing resources on-demand. Rather than building data centers that can serve peak capacity for all of their applications, they can instead build data centers (private clouds) that serve the typical capacity of their mission-critical applications and then overflow to external clouds as necessary.

But to support this, applications must understand the cloud computing model and be able to scale across internal (private) and external clouds.

The next big opportunity is around interoperability and portability across the many cloud services out there (Amazon, Google, Rackspace, etc.) so customers can consume services without fear of lock-in and the high costs of being dependent on one vendor. This is where legacy multi-tenant SaaS applications like SalesForce begin to fail. They run in only the vendor’s data centers, not across public and private clouds. They are unable to take advantage of cloud computing.

This assertion would seem, at face value and somewhat unintentionally, to be backed up by a slide that Phil Wainewright used in a recent Whitepaper, Redefining software platforms — How PaaS changes the game for ISVs (the writing of which, by way of disclosure for both Phil and myself, was supported by the Intuit Partner Platform who is a client of both of ours). The diagram (see below) shows the increasing duplication of infrastructure, and hence increasing inefficiency as one moves from shared infrastructure, to dedicated SaaS and through to on-premise.


It’s a powerful diagram and one which, by extension supports Augustin’s contention.

Jumping into the sphere of the discussion, if a little peripherally, comes a survey from Avanade. Hat tip to GigaOm for finding it and making the comment that the survey found that;

By a ratio of 4:1 (2:1 on a worldwide basis), respondents said they would prefer to have their applications delivered as services from internal platforms. Is SaaS the “killer app” for internal clouds?

But I came away wondering if Augustin was looking at the issue completely as a binary argument, when clearly it is not. It’s a reasonably well accepted, if seldom mentioned, fact among the clouderati that large SaaS vendors often offer their products to large enterprise customers as “private SaaS” that is SaaS applications installed on private datacenters where organizations can feel better about their qualms over security, reliability and all the other barriers corporate IT puts up to justify apprehension about SaaS.

So like many of these things, it’s very much a question of horses for courses. In the same way that Augustin’s own product, SugarCRM is available on-premise, via IaaS or any combination of the above, so to will other vendors slice and dice their products to meet the particular requirements of their customer base. For the SMB set, much more important than the on-premise/on-demand debate is the ability to utilize web apps with their existing workflows and product choices – this is where offerings like the IPP with it’s built in connection to QuickBooks on the desktop comes in. For larger customers the imperatives are different, security and reliability are top concern and, whether rightly or wrongly, their perception is that internal infrastructure gives them the best results over these metrics.

So the takeaway here is one of flexibility – as an industry, and given the nascent nature of cloud computing and it’s triumvirate of categories – IaaS, PaaS and SaaS – we do ourselves a disservice by being too dogmatic about these things.

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.

  • Totally agree about not being too dogmatic. At Central Desktop we run on our own private multi-tenant SaaS platform. Essentially, our private cloud.

    Over the years, we’ve continued to evaluate Clouds as an option for delivering point services to our customers – but the economics havn’t justified a switch.

    What is often overlooked by SaaS pundits is the fact that connecting to a Cloud very often reduces (sometimes eliminates) key features that a private SaaS vendor can provide. For example, we looked hard at augmenting our file storage with cloud storage via Amazon (and a few other vendors) but decided against it for two main reasons:

    1. Vendor Economics. Yes, economics. The transactional access of files and data that our customers require as active collaborators was a very expensive alternative. Remember, we are talking about Terabytes and Terabytes of data that we host for our clients. Sure, Cloud storage/services are cost-effective for nascent startups trying to save money ( I get it…I really do), but onec you develop into a serious business with lots of paying customers and transactional needs, Cloud becomes more cost prohibitive than one might imagine.

    2. Feature Reduction. Moving hot data to the Cloud resulted in slower performance and a reduction of features such as the inability to easily provide full-text search for documents stored in an offsite Cloud. The bar to provide certain features “in the cloud” was more expensive than it was for us to “do it ourselves.”

    I agree that we shouldn’t be dogmatic, but the economics of cloud storage and delivery have been prohibitive to us and our customers for the reasons listed above.

    As a pseudo-cloud pundit myself, I’d also argue that the “open platform” Cloud services promises customers is still hedged with the “single point of failures” that we continue to see in Cloud vendors (virtually all of them have experienced extended outages with their vendors left wondering “what is going on?”).

    At least when we go down…..we know why, we are in control of the situation and we can fix it ourselves……and we can tell our customers what is happening/going on. Luckily, we don’t go down very often at all: http://status.centraldesktop.com

    Will we someday surpass the ability to efficiently and cheaply deliver our SaaS’s to our customers on our private stack? Perhaps, but we’ve also found that our homegrown architecture enables lower cost of goods and is becoming more and more of a competitive advantage when we look at many of our competitors.

    Long live profitability!

  • Issac – great response… thanks a bunch. Well reasoned and rational! well done!

Leave a Reply

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