Cloud platforms for profit

DFN: This is is kind of the same ‘dilema’ in telecommunications, facilities based network or ip based network. I’m interested in traditional software (ie Essbase on my own servers $2M per installation) versus Saas ($200K per installation) for potential future deployments of budgeting & forecasting systems / processes.

Guest View: Cloud platforms for profit
By Peter Coffee

November 15, 2009 —
Because it’s always the first question that people ask, let me answer it up front. Yes, I’m the same guy who wrote those developer tool reviews, and developer-centered columns, in "PC Week" and "eWeek" (as well as on from 1989 through 2007. Yes, I’m now with, not merely lending them my name, but working there for what’s now quickly nearing three years.

My journey from the cushy VIP skybox of the technology trade press, down to the windy (and often muddy) playing field of the IT marketplace, was driven by forces that I’d like to take this chance to describe to fellow developers (yes once more, I used to write code for a living). I believe that these forces will soon compel every developer to think about a once-in-a-generation shift, already well under way, in what developers will get paid best to do—and in how they’ll do it.

Stretching the timeline
Toward the end of my time at "eWeek," I found that I was generating many fewer reviews of development tools than I had in my earlier years. An evaluation timeline that had once required only a morning to get a tool installed, and to build at least a few functional demonstration applications, had stretched into a process that took several days merely to get a new tool wired into the cumbersome stack of data resources and middleware layers that typify modern business app development.

If it took me this long to do anything, I wondered, what was it like in an enterprise environment, where a new tool had to become a robustly integrated part of everything?

I saw the beginning of a better way when I saw Web-based applications evolving into open application platforms. I don’t mean “open source,” at least not necessarily, but rather “open to innovation”: platforms on which you could define a domain-specific data model, devise custom logic to wrap a business process around that data, and design a user experience to turn that business logic into an approachable and convenient tool.

The question, of course, was whether developers and the people who pay them would ever accept the idea that a valuable business application—perhaps a crown jewel of intellectual property—would ever be placed on someone else’s systems, running under someone else’s administration, being provisioned back to its authors or to its authors’ customers as a service. Was the developer marketplace ready for that change of the game?

It seemed to me that the tipping point had come when I saw Web-based platforms—yes, I believe that the P-word has meaning, whatever Dilbert’s colleagues may say—handling everything from finances to healthcare to personal productivity. I saw both individuals and enterprises using a panoply of Web-based applications, including online discount stock trading, online prescription management, collaborative word processors and spreadsheets, and increasingly personal tools such as Facebook.

To anyone under 20 years old today, the main function of a PC or a Mac is to run a Web browser, and to do the things that are thereby made available.

Meanwhile, in enterprise and industry, Web-delivered infrastructures (such as Amazon’s Elastic Compute Cloud) showed that the cost-effectiveness of computing at the other end of an Internet wire was quickly outgrowing older ideas of “outsourcing.” Computing was becoming an elastically consumed and priced commodity: a concept that’s widely celebrated, though substantially (even if necessarily) over-simplified, by popular treatments like Nicholas Carr’s "The Big Switch."

I also saw an applications marketplace that was no longer demarcated by on-premise versus Web-based delivery. The novelty of “Software as a Service” faded quickly; the major bifurcation of business models became the choice of supporting a “free” Web application with online advertising, or selling actual subscriptions and offering professional support and commercial-grade levels of service. Trying to start up a new software venture that was not Web-based was coming to be a waste of time: No one was interested in funding such an effort.

Into the cloud
And then came the label of “the cloud.” Worldwide search traffic for the phrase “cloud computing” has grown by roughly a factor of six in the past 18 months, as measured, with ironic self-referentiality, by the cloud-based tools of Google Trends.

This surge of interest coincided, but is not merely coincidental, with a time when IT budget growth has not merely slowed, but has actually gone negative. We’re well into the first IT spending contraction since 2003, as measured by Goldman Sachs.

Perversely, though, the same people who have less to spend are being required to do more. They’re impelled to meet rising expectations for richness and availability of customer self-service and partner integration capabilities, for which there are at least good business-case justifications. They’re also under mandates to meet more stringent demands for information security and business process governance, for which there’s no real ROI, but these are clearly requirements for staying in the game.

Those application buyers are looking to the “cloud computing” model to give them a combination of lower up-front capital investment, radically faster deployment, dramatic reduction of technical risk, and strategic opportunities to experiment at low cost—and to scale-up successful experiments at previously unthinkable speed.

That’s the demand side of what’s driving development into the cloud. What about the supply-side forces?

SUBHED: What’s in the cloud for developers?
For all of the reasons already described, application developers should be looking to the cloud as the focus for their development of new skills, and for their thinking about the opportunities that they would like to pursue for personal achievement and professional success. The good news is that developers will also be having a lot more fun, making a lot more money, and getting to advance the state of their art at a far more rapid pace than they have during the past twenty years.

When you build applications for a traditional IT environment, you have to begin with platform-centric market research. What is the preferred operating system among your target customers? What database engines must you support? Which middleware stacks must you complement? Bor-ing—and, by definition, exclusionary. You have to decide which customers you’re not going to serve.

Here you are, with some genuine inspirations as to how a complex task can be made simpler, or an exotic knowledge base can be put to work for more people. Patience, grasshopper: You’ll get to do some problem-centric work in another few months…

…but then you’ll have to find some way to rise above the noise floor of the marketplace, getting the attention of people who are used to buying applications from a shockingly tiny handful of vendors.

Gartner, in particular, projects a global enterprise software revenue stream of US$222 billion during 2009, but SAP, Oracle and Microsoft alone will take in roughly half of that figure (albeit not all of that in the enterprise marketplace). How do you achieve credibility simultaneously, as an enterprise application provider of proven readiness for prime time, while also offering something truly new?

Here’s one way to do it. Don’t write the application as software that has to be sold (excuse me, licensed) to customers who must then find infrastructure on which to run the code. Don’t make a huge investment of your own in a massive data center that can handle (what you hope will be) the fast-growing demands of a hungry marketplace to consume your application as a service.

Go cloudwards. Go out and identify the most productive environment in which you can do what you believe you’ll do better than anyone else. It does not matter (repeat: It. Does. Not. Matter.) whether the environment you like is popular. It only matters whether you can build something great in that environment, and then make that work accessible to customers over standard Internet protocols and connections.

Want to write it in Lisp? No problem. Want to run it on Linux? No problem. Can you support the major Web browsers as your client layers on your prospective customers’ endpoint devices? Almost certainly. So start doing something wonderful.

Depending on your target market, you’ll choose your cloud platform based on any of various criteria. For consumer-facing offerings, you’ll favor big bandwidth at low cost. For enterprise-oriented offerings, you’ll emphasize high uptime and a long list of security certifications. For compute-intensive offerings with self-contained data worlds, such as gaming, you’ll look for low price per CPU cycle. For data-intensive offerings that derive huge value from integration with everything, you’ll look for cloud platforms (such as Salesforce’s, if I may get away with one tiny bit of employer self-promotion) that are broadly supported by integration tools and expertise.

Whichever platform you choose, you’ll bring the track record of that Platform-as-a-Service forward into the marketplace as part of what you build upon it. If that platform has a reputation for high uptime, a solid list of audits and certifications for the strength and the integrity of its security apparatus and practices, and a thriving partner ecosystem of people who know how to integrate other IT assets with that platform’s capabilities, then welcome to the club. Your brand new application now shares in those aspects of maturity.

On beyond “bits in a baggie”
I remember the early days of the IBM PC, or even the Apple II before that, when being an independent software developer meant bulk-copying floppy disks that your local computer shop would sell from a hookboard display—about as high-end a market as selling novelty toys at the local Party City store.

No matter how much nicer the packaging was, or how much higher the prices were, the software marketplace B.C. (Before Cloud) had not really evolved in any fundamental way. It was still, essentially, bits in a baggie (a trademark of Pactiv Corp.), and in the eyes of many established software providers, that remains the way that they’d prefer to deal with their own customers and with the customers of their developer ecosystem.

It’s an old way, but no longer the best way, and increasingly it’s clear that it’s not tomorrow’s way.

Development can be more:

• Developers can demand diverse tool sets and environments that suits the diversity of the problems to be solved, without the mediocritizing mass of a world full of Wintel boxes forcing everyone into the same development workshop.

• Developers can stay on the cutting edge of creating new capabilities, instead of devoting the majority of their efforts to maintenance of their installed base in a worldwide spilled ants’ nest of slightly different on-site installations.

• Developers can add value by constantly improving their offerings, instead of reining in their enthusiasm so as to avoid burdening their customers with the nuisance of testing and deploying frequent upgrades. Cloud-based platforms can make value-adding upgrades transparent, inexpensive and low-risk.

In the cloud, you can use the tools that do the most for you. You can put your effort into doing what makes you special. You can make it better, all the time, and have your customers love you for it.

I know where I want to do my work from now on, and I hope to see you there.

Peter Coffee, a former columnist for PC Week and eWeek, is now Director of Platform Research for


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: