Kirk Klasson

Larry, you evil genius…

A master redefines cloud computing without even trying…

I’m not Larry Ellison’s biggest fan but I admit he has more of a Midas touch than most in the technology industry. His acquisitions are usually behind the curve, often buying customers rather than technology, and his innovations are much the same gaining as many patents through acquisition as invention but he has a knack for making the most out of what he’s got. This week he put the cloud industry on notice that while he may not be the prettiest girl at the prom, showing up late and in a borrowed dress, he might be the one you’ll want to take home.

Oracle is a funny place. For years it has tried to rationalize its application strategy through a larger vision called Fusion which the market has variously failed to hear, recognize, comprehend or acknowledge, take your pick. Not entirely unanticipated since Fusion was competing with another vision from Oracle called Applications Unlimited. But that doesn’t really matter since Oracle is not really as much a tech company as an operation that Tony Soprano might envy. Oracle really doesn’t sell you technology as much as buy your existing technology contracts and then usuriously insinuate what your life might be like in the absence of Oracle technology which is exactly what Larry did this past week out in OpenWorld in San Francisco, standing in front of what I imagined were pallets loaded with bags of cement.

For the past several years industry pundits have been belittling Oracle’s cloud offerings as not genuinely cloud based, not architecturally pure, a mish mash and rehash of tired SaaS offerings masquerading as something they are assuredly not, sweetly scented, dewy clouds. And if you accepted their premise you might deem this to be a credible assessment. Oracle’s customers have been reticent about directly investing in Fusion, relying instead on Oracle to incorporate service orientation into existing SaaS products. Similarly, Oracle has maintain a very proprietary position when it comes to cloud delivery systems keeping open source initiatives embraced by other vendors at arms length while promoting cloud PaaS and IaaS services that only operate with Fusion Middleware. Not a very customer friendly posture if you’re assuming that customers want to own this portion of network based application delivery. But, in spite of all of these apparent shortcomings, Oracle may yet emerge as a trusted cloud partner not because of their technology portfolio but rather due to simple customer realities.

Come fly with me, let’s fly, let’s fly away…

Purists in the cloud biz, and there are many, mostly come from the Amazon Web Services (AWS) school of architectural piety. One has to go all the way back to 2006 and the introduction of AWS to understand how this came about. Amazon’s or more correctly Werner Vogels’, the Werner Van Braun of web services, vision was based on principles that emerged from Service Oriented Architectures and Grid based computing techniques. It is from these methods that cloud based services derive their primary benefits of agility, scalability and robust recovery. Werner was trying to solve several issues at once that were mostly unique to applications that were exclusively web and browser based, had highly variable transaction volume and intensity but relatively low degrees of complexity and functionality in that all requests could be satisfied with a limited number of common core software based services.

Architecturally the preferred way to address these requirements was through a series of loosely coupled, network connected software components where demands could be easily parsed, distributed and executed by numerous compute facilities under common transaction orchestration. Things had to scale out, replicated services over numerous network connected commodity resources, as opposed to scale up, infinite amounts of processing and memory capacity inside a single computing facility. And for architectural purists and particular types of applications, usually web based, mobile, grid or those that include a large share of content distribution, embedded services or variable transaction intensity, this represents a very elegant and compelling solution.

However, in the overall scheme of things, cloud based web services represent a special purpose architecture (see Requiem for a Business Model…January 2011), one that may in time come to dominate enterprise computing, but not necessary the only one or the one best suited for all applications. At this point, customer facing, transaction intensive, greenfield initiatives are especially good candidates for a clean sheet re-write targeting cloud based web service architectures hosted on merchant based multi-tenant infrastructure. For any given large enterprise this might represent 10-20 percent of their application portfolio. Beyond that, the selection criteria for applications suited for pure cloud, merchant based architecture gets a little blurry and the realizable rate of return from application migration gets a bit sketchy. It may well be that over the next several years cloud services become the preferred swing space in a larger architectural transition to distributed web based applications. However that may not be your only or preferred option and Larry pointed this out by saying that Oracle will move you to cloud services if you want but, more importantly, would move you back. The implication was simple but profound, your application might be better suited to a more mundane general purpose architecture and, if so, you can spare yourself all the cost and headaches of having to re-write your applications to attain what might be marginal benefits from a pure, cloud based, special purpose architecture. And with that simple suggestion, Larry pulled on the most obvious loose thread in the temple of cloud-dom, and Superman’s cape began to unravel.

… if you could use some exotic booze…

During the past several years the benefits of cloud based computing, agility, security, scalability, resilience, have become better understood, at the same time the avenues to achieve these benefits have also come into clearer focus. Most of these benefits for existing applications can be realized through cloud emulation techniques such as desktop and server virtualization without the need for an extensive re-write of the underlying code or scraping the assets they run on. Cloud based emulation also comes with a cost and it becomes important to be able to compare these costs to the costs and benefits that might be derived from a complete cloud based application re-write.  But unlike in prior architectural transitions such as mainframe to client/server it is not altogether clear why users should bear the brunt of moving to cloud based architectures (see Forecast, Partly Cloudy…July 2010). The assumption is that the economics of multi-tenant, cloud based services, white box hardware and open sourced software will entice users to make the trip. But the truth is, transaction variability aside, those same cost advantages should be available to users who want to simply refresh their proprietary environments and build out private clouds solutions when and where it suits them.

The bottom line is that most users are trying to reduce the amount of complexity they have to assume when evaluating the network based delivery of applications, whether that be pure, merchant- based clouds, emulated clouds or SaaS and they are likely to employ some or all of these options at some point in the future. This became evident when a few days after Larry’s address at OpenWorld, Stephen Little the CIO of Xerox was quoted in the WSJ as acknowledging the potential cost benefits from pure cloud infrastructure but indicated he was reluctant to move beyond his current SaaS providers due the complexity of integration that he was likely to encounter. This suggests that the integration of infrastructure options is looming larger in the minds of enterprise users than the selection of pure cloud architectures. And if you consider this for a moment, it makes a lot of sense. How many organizations, when evaluating SaaS, insist on a thorough architectural review of the vendor’s code to determine whether or not the application in question was written using the most up to date REST based, map-reduced, cloud conforming principles? How many Oracle customers examine their SaaS applications to determine to what degree Fusion has been consummately incorporated?

My guess is zero because at bottom who cares?

…there’s a bar in far Bombay

So, having climbed out on this limb, purveyors of merchant-based, multi-tenant, pure cloud computing services are left with an interesting dilemma. Enterprise users are waking up to the fact that merchant delivered cloud computing is not some new tsunami but rather just another option and one that they can and should evaluate as part of a portfolio of computing options, including proprietary management of existing systems, cloud emulation, private clouds, SaaS, etc. Some suppliers like Amazon or Google whose business models don’t necessarily reflect the impact of cloud services and view it more as a cost dilution rather than revenue contribution may not care. However, for legacy technology players with established on-prem infrastructure businesses, including the likes of HP, IBM and yes, even Oracle, the rush to substitute existing revenue streams for cloud based revenues, and, let’s face it, for most of these vendors that’s what this amounts to, the stakes are high and about to be raised.

Moving forward there won’t be a user worth their salt that won’t obtain a PaaS or IaaS quote and compare that to what they are currently doing and try to reconcile the differences along with weighing the risks and costs of re-writing their application code and learning how to work with a whole bunch of new open-sourced or proprietary management consoles (see Here We Go Again…April 2013). Further, most users have concluded, as Larry just suggested, that their options are likely to change so swallowing a bunch of transition costs at this point along with scraping their existing computing platform is probably not a prudent course of action regardless of how sunk those costs might appear up against Google’s latest price cut.

And the conversations between suppliers and buyers are bound to get interesting. So, let’s see, I can use your multi-tenant cloud services for less than a penny an hour compared to the same configuration of your gear, software and services which you sold me two years ago that are amortizing at $100 an hour down at the data center….. and I thought we were friends. And who might benefit from these conversations? These closer evaluations? From users taking their time or even standing still? The same shop that stayed out of the cloud game and preserved their margins for the past 6 years. That’s right, Oracle.

So once again, as our graphic clearly indicates, Larry has shown the entire industry exactly what kind of electricity you can generate with simple customer realities, a practical premise and a large pair of balls.

Genius, pure genius.


Graphic courtesy of the Boston, Museum of Science

Sorry, the comment form is closed at this time.

Insights on Technology and Strategy