Coming Down from the Cloud
By David Diamond • Oct 15, 2014
When I first met Picturepark founder Bruno Jehle, some 15 years ago, he told me, in one of the detailed, animated predictions for which he’s known, how Application Service Providing (ASP) would change the perception of IT and the dynamics of the software economy in general. Shortly thereafter, as the burst of the dot-com bubble buried many dreams and promises, Bruno saw his vision realized: ASP became Software-as-a-Service (SaaS) and lead to the mega trend we today simply call “The Cloud”.
What a visionary!
For more than ten years since its birth in 1997, Picturepark was available only as hosted SaaS. At the time I joined the company in 2004, we turned down virtually every sale that required on-premise installations. And the few exceptions we made didn’t pay out.
I was always impressed at how well the Picturepark Cloud made for a win-win situation for the primary user, marketing and communications, and the approvers, IT. Marketing liked being served from an external provider, without being held back by IT project planning that could, itself, be held back by limited resources. And because these bulky digital assets pestered IT’s expensive, mission-critical infrastructure, IT were rarely unhappy to see these exotic DAMs hosted externally. And both sides appreciated that users required only Web browsers to use Picturepark, so no client or server applications were needed.
The subscription model with lower payments made it additionally easy for marketing managers to run DAM on a discretionary budget, even swiping their credit cards instead of budgeting and approval-hunting for expensive up-front enterprise software purchases.
On-premise or Off the Table
While the world started to buzz about Cloud-provisioned software being the sole future place to be, we realized that it might be only part of our own success story.
While subscriptions to the Picturepark Cloud increased, demand for Picturepark on-premise installations increased too. Reasons ranged from data being too sensitive to cross borders or become visible to the wrong eyes, to the need for 24/7 guaranteed accessibility, to the need for high-speed data transfers, to the need for same-host or same-data center integration, which wasn’t possible via the Cloud.
This was no surprise to us—we had heard these rationales many times. But as the volume of instance increased, the idea of Picturepark on-premise became increasingly attractive. Still, it seemed a bit of a backward move for a company that can reasonably claim to be one of the pioneers of SaaS. We needed further convincing.
We conducted some research to better understand not just the business rationales for Cloud vs. on-premise, but the psychological aspects too. Why were people afraid of the Cloud? Or why were people so convinced that on-premise would never go away? And even more interestingly for us, why were some people so quick to give up on on-premise installations? After all, every Cloud is on-premise somewhere.
Almost immediately we realized that the pro-Cloud mantra, “it requires no infrastructure, meaning less capital outlay”, was increasingly less a message point of the on-premise naysayers. The reason was that the line between on-premise and Cloud was blurring for everyone, including IT.
IT Discovers the Customer
The new Cloud Economics had IT departments under pressure, forcing them to reinvent themselves, sometimes entirely. For instance, instead of just sticking with the best-of-the-best in computer hardware, IT started adding other tiers for less business critical applications. In many cases, IT began to rely on external Cloud services to manage their on-premise infrastructures. In time, the smarter IT departments became much more service oriented themselves. On the other end of the network connection now was a customer, not just a user.
Not to be left out of the Cloud advantage, IT started outsourcing internal infrastructures to regional hosting infrastructure providers, who had built businesses that mirrored what the global Amazons, Googles and Microsofts had built.
Many of these smaller players were no newbies to hosting, but survivors of Old School providers who started their businesses twenty years before. Facing increasing global competition, their choice was to either broker AWS or Azure services, focus on a niche or, of course, fail as a business.
Computer virtualization and other technologies enabled them to compete on price, while offering scalable infrastructure like the commodity Cloud infrastructure vendors. But this alone wouldn’t be enough. They had to provide something the Cloud giants couldn’t touch.
Governments Discover the Cloud
As governments started paying attention to data, they indirectly provided niche infrastructure providers with long-term life support. Cloud giants are willing to discuss custom SLAs and tailored legal agreements with other Fortune 1000 giants only. But scalability is based on standardization, so they can’t offer unique terms and conditions to everyone and still expect to scale efficiently.
Niche providers, on the other hand, were able to negotiate contracts because, of course, they had far fewer contract to consider. From this, a new type of Cloud was born—the private Cloud, the specialized Cloud or, as Picturepark calls it, the Regional Cloud.
As an alternative to staying on-premise or compromising to work with the commodity public Cloud giants, organizations in need of IT infrastructure had a third option: defining their own terms and working with the data center down the street.
Picturepark Falls to Earth
Upon realizing this trend, concluding what we had to do was a no-brainer: We not only had to make Picturepark work on-premise, we had to reengineer Picturepark so that a single installation could serve a single organization or a thousand organizations. We had to give each instance of Picturepark the ability to become a DAM Cloud of its own.
Frankly, this was a major architectural challenge that took us longer than we expected, and cost us quite a bit of money. When you build Cloud software, you don’t need to think about things like installer scripts or back-end administration utilities, and you need to make the software work in only one server environment—your own. Further, when you find bugs, you can fix them without having to think about distributing software updates tested for all the differences you’ll find in other IT environments.
In truth, the SaaS model is very attractive (and lucrative) for software developers, as is evidenced by larger companies like Adobe and Microsoft headed in that direction.
The challenge for us was to build a reliable, portable version of Picturepark, while not slowing the speed of development we had with our Cloud system. Our goal was one Picturepark that could do it all—no multiple code bases and no user experience differences.
So that’s what we did: Not a single line of code is different between same versions of Picturepark Onsite and Picturepark Cloud. With the proper infrastructure and license in place, each Picturepark Onsite installation can serve a DAM Cloud of its own, with multiple instances running. In fact, our global Picturepark Cloud software, on which we’ve been running hundreds of customer instances for well over a decade, is no different than the Picturepark Onsite software installed and in use by our on-premise customers.
This architecture enables us to also scale globally while remaining local by offering Picturepark Regional Clouds. Hosted by specially certified partners, a Picturepark Regional Cloud is like a copy of our Global Cloud infrastructure that comes with local services, is sold in local currencies, can offer custom SLAs and warranties, and is housed entirely within the clear boundaries of local legislation. It’s a niche cloud of its own.
PaaS the Handcuffs
Between the raw computing provided by infrastructure-as-a-service (IaaS), and the turnkey readiness of software-as-a-service (SaaS), lies a middle ground known as platform-as-a-service (PaaS).
While PaaS is a new buzz-term mainly coined by Microsoft Azure, it has existed for a while in the form of specialized services, such as the Amazon Web Services S3, a proprietary storage service, Amazon DynamoDB, a proprietary NoSQL database service, or Google App Engine, a proprietary app hosting environment.
Software developers that offer their solutions solely in the Cloud are strongly incentivized to use these PaaS technologies because they can benefit from available and auto-scaling infrastructure and managed services, without having to maintain even (virtual) storage or database servers. In other words, they can build the application on, for example, the AWS “operating system”.
But the drawback of this approach is proprietary Cloud technology and increased switching costs: The software becomes locked into the chosen Cloud platform and is never again likely be able to run anywhere else. Once AWS, always AWS.
This is not a problem if you believe that only the Cloud will survive, and that anything on-premise is an old-economy option for laggards ultimately doomed to fail. But between on-premise and commodity Clouds, the new niche Cloud options will grow, and they will require software that installs and works in those Clouds just as it works on-premise.
Too bad if you can’t offer a truly scalable Cloud solution with one of the commodity cloud providers; but too bad also if that same piece of software can’t be deployed on-premise. Cloud “lock-in” might seem like a nonissue today, but regulatory issues surrounding data storage are not going to let up any time soon. Privacy laws that went into effect in Australia in early 2014 demonstrate that this is more than just a theoretical nuisance. These laws were, in fact, the reason a Picturepark Regional Cloud was deployed by our Australian partner, DataBasics.
Additionally, the need for standardization of the public Cloud providers will prevent them from offering the features, specialties or trusted professional services some customers require. Softlayer, for instance, specializes in “bare-metal options” for better speed and performance, and provides a huge API for full control, required for highly mission-critical operations. No surprise that IBM acquired the company last year. Other samples of fast growing companies are Firehost, which offers specialized and high-end secured hosting services, and CloudSigma, with high-performance in storage and computing at a better price than any of the public Cloud providers.
In retrospect, I couldn’t think of a better situation to be in with regard to deployment than we are today, and for the future. Our prospects and customers don’t have to decide between Picturepark and another DAM just because of Cloud or on-premise requirements. They have the option to host their data with the highest scalability in our standardized global Picturepark Cloud, with one of our certified Picturepark Regional Cloud providers, or to host it on-premise or with any other Cloud infrastructure provider of choice.
Even better, Picturepark customers have the option to build their own private Picturepark DAM Clouds and serve multiple departments or clients, leveraging the very same tools and technology we use to manage global DAM Clouds for large global enterprises.
Bruno was right: It’s all about service orientation, which ultimately means remaining customer-focused and providing them a choice.