Saturday, March 02, 2013

Cloud Confusion on The Motley Fool

As I've written in the past, financial analysts may provide good advice for investors in the tech sector. But their analysis is not very useful to buyers of technology products and services. It's not that they don't have insights, but they are writing for a different audience: investors, not customers and prospects.

Some parts of the financial press are another story. Some financial media reporters so poorly understand the tech industry that neither investors nor prospective buyers should listen to them.  

I saw an example of this today on The Motley Fool, in a story entitled, Is Oracle's Cloud Really Fake? In it the contributor, Richard Saintvilus, takes issue with an Infoworld article by David Linthicum that criticizes Oracle's most recent "cloud" announcement as "faux IaaS."

The Motley Fool is a website aimed at the individual small investor. It provides both free content as well as paid subscriber material. It also makes money from advertising. Therefore, generating page views is a key objective, with much of its content generated by freelancers, as appears to be the case here. So, the quality of its content varies.

Apologies to in advance to Saintvilus, who reached out to me on Twitter after I sniped at his story. He asked for specifics, so here you go.

Oracle Late to the Cloud

Staintvilus immediately starts with a misconception, that Oracle was an early proponent of cloud computing. Here is his lead:
Wall Street loves a hot trend and had essentially decided four years ago that the cloud was next. With corporations needing all of the cost savings/productivity benefits the cloud offered, the timing was perfect. Oracle was one of the first blue chip enterprise companies to realize this opportunity. [emphasis here, and throughout, is mine.]
Sorry Richard. Even as late as 2009, Oracle's CEO Larry Ellison famously mocked the term cloud computing, calling it no more than a fashion statement. Furthermore, he not only mocked the term, he mocked the concept. To this day, Ellison criticizes multitenant applications, which are the cornerstone of most large scale SaaS providers. Only recently, as Oracle realized it was losing the war did Oracle embrace cloud computing, albeit with its own twist (either hosted single tenant applications, or multi-tenant applications with single tenant databases.) Industry analysts can argue all day about the relative merits of each approach. But none would claim that Oracle was anywhere to be found at the beginning of the trend to cloud computing.

Oracle Market Share in Cloud Services

He continues with a comment on Oracle's rising revenues:
The company [Oracle] is providing a service, of which there has been very few complaints -- at least not according to the rising revenues, which suggest it's stealing share from rivals such as salesforce.com. And Oracle is not the cheapest on the market, either. So, there's a reason why customers are willing to pay the premium. And it's not because these corporate CIOs are dumb.
Yes, Oracle's revenues are rising. But those of Salesforce.com are rising also, up 37% in 2012. Furthermore, while all of the revenues of SFDC are derived from cloud computing, only a small percentage of Oracle's are. So to impute on the basis of revenues that Oracle is taking market share from Salesforce.com is ludicrous. Certainly, in my own firm's work advising buyers in software selection, I do not see Oracle taking market share from Salesforce.com. In fact, in CRM, I see deals in which buyers want to look at Salesforce but do not even consider Oracle, especially in the midmarket.

As far as software pricing is concerned, neither do I see Oracle as commanding a premium price over Salesforce.com, or over other application vendors for that matter. In fact, Oracle is notorious for competing on price when it really wants a competitive deal, as it knows it can make it up on maintenance revenue in the future.

But the author wouldn't know that, because he does not advise technology buyers, nor is he in a position to see actual deals going down.

Oracle's Engineered Systems Are Not "Cloud"

He then confuses Oracle's new Exa-boxes with cloud services.
However, David Linthicum of InfoWorld thinks [CIOs are] idiots (I'm paraphrasing that a bit), which doesn't make sense. CIOs are spending billions annually with Oracle. But in Linthicum's recent article, he insists they don't know what they're buying: "Oracle is continuing its faux cloud strategy, adding to its private-cloud infrastructure offering the ability to rent for a monthly fee preconfigured application servers to be deployed in customer data centers. The available application servers -- what Oracle calls 'engineered systems'"
The author can be forgiven for this misunderstanding, as Oracle itself confuses this issue, which is the whole point of Linthicum's criticism. The basic point is that cloud computing is a "service," whereas Oracle's computer hardware (whether old school Sun commodity servers, or Oracle's new "engineered systems") is a physical product. Renting Oracle hardware does not magically turn the hardware into a cloud service.

Oracle's Engineered Systems Are an Old Concept

He continues with the impression that Oracle's Exa-boxes are somehow a new concept. 
Linthicum clearly has an ax to grind. While he's going all-out on Oracle's product portfolio, rival companies have been working hard to duplicate Oracle's offerings. For instance, IBM has a rival offering called PureSystems -- launched three years after Oracle's Engineered Systems, or ES. And, after Oracle has already deployed ES to more than 1,000 customers in 43 countries, IBM followed. Big Blue has gained traction, but not to the extent of Oracle. And I doubt that IBM would have followed a model it didn't think had sustaining potential.
The author appears unaware that Oracle is attempting to return to the IBM era of the 1960s. In fact, Larry Ellison has said so himself. The IBM mainframe at the time was a single integrated platform of hardware, operating system, database, and applications engineered from the ground up to work together. IBM's AS/400 series of machines (now called Series i) took this concept even further. What broke up IBM's dominance in the mainframe era was the fact that these boxes were all based on proprietary standards, and eventually low cost commodity hardware (whether IBM personal computers, or later, Unix boxes from providers such as Sun) could do the job much more cost effectively. The downside was that the new approach led to challenges in system integration, in making all the layers of the technology stack work together.

Oracle's strategy with its Exa-boxes is to return to this single technology stack from hardware through applications, engineered from the ground up to work together. Will Oracle be successful? Only time will tell. And, certainly the 1,000 customer number that the author mentions is not yet enough of a measure of success.

Regardless, what does any of this have to do with cloud computing? Absolutely nothing. Oracle is launching a public cloud offering, with its Exa-boxes as part of the infrastructure. Other providers can do the same using commodity hardware, as Google, Amazon, Microsoft and others have already done.

But the Oracle offering that Linthicum is criticizing and that the author is defending is not Oracle's public cloud service. Rather it is an arrangement whereby Oracle customers can, for a monthly fee,  rent preconfigured Oracle application servers and run them in their own data centers. Linthicum is absolutely right: this has nothing to do with cloud computing.

According to the NIST definition of cloud computing, there are five essential characteristics of cloud computing, and Oracle's hardware rental offering does not satisfy four of them. (See the link in this paragraph for a more complete definition.)
  • On-demand self-service. Oracle's rental agreement does not allow the customer to unilaterally provision computing capabilities, such as server time and network storage, as needed automatically without requiring human interaction with each service provider.
  • Resource pooling. Oracle's rental agreement does not pool computing resources for multiple customers in a multi-tenant model, with different physical and virtual resources dynamically assigned and reassigned according to consumer demand.
  • Rapid elasticity. Oracle's offering does not allow computing capabilities to be elastically provisioned and released, in some cases automatically, to scale rapidly outward and inward commensurate with demand. As Linthicum points out, the customer has to pay extra for spikes in demand and there is no provision to ramp down demand, and cost.
  • Measured service. Oracle's offering does not automatically control and optimize resource use by leveraging a metering capability. Customers do not pay as they go.  
Clearly then, Oracle's offering may use the terminology of cloud computing but it does not display the essential characteristics of cloud computing. You can call me a fish, but that doesn't make me one.

The List Goes On

I have neither the time nor the patience to go much further. Let me just list a few (and by no means all) of the remaining errors.
  • "Linthicum is pretending to be an expert on something that is still in its infancy." Cloud computing may not be a full grown adult, but it is certainly not an infant. Providers such as Salesforce.com have been delivering cloud services for well over a decade, ancient history in the technology industry.  
  • "Oracle innovates at the technology layer, thereby giving customers more leverage and independence from consulting fees." What exactly is the "technology layer?" Everything from bare metal hardware to business applications are "technology." Furthermore, talk to Oracle customers: I doubt anyone will tell you they have less need for consultants. 
  • "Had Cisco contracted out its cloud services to Oracle, it could have remained focus on growing its business." The types of cloud services that Cisco offers, such as cloud network management, are not services that Oracle provides. Therefore, it would not be possible for Cisco to contract with Oracle to provide those services on behalf of Cisco. 
  • "Even if Linthicum's pricing claims were correct, then it means Microsoft's Azure, which has a pay-as-you-go model, is also fake. But Microsoft has been reducing its prices because it can't compete." If Microsoft is lowering its Azure pricing, it's not because it can't compete, but because it can deliver cloud services at lower and lower costs over time, as it scales and the costs of technology drop (see "Moore's Law"). Similarly, Amazon lowers its AWS prices multiple times per year and no one in his right mind claims it's because Amazon can't compete.
Parts of The Motley Fool article are nearly indecipherable, especially toward the end. But I think this is enough to illustrate: parts of the financial press are poor sources of information on enterprise IT. By that, I do not mean to imply that all financial reporters are suspect. One would not expect to see a story such as this one to appear, say, on the pages of The Wall Street Journal or Financial Times.

Furthermore, none of my criticism should be taken to mean that Oracle is not a good company from either the investor perspective or customer perspective. As the author tweeted me, "Oracle is one of the best tech names on the market and it deserves fairness."

Yes it is, and yes it does. And because of that, it also deserves accurate reporting.  

Related posts

Enterprise IT Buyers: Don’t Listen to Financial Analysts
Oracle's Behavior Undercuts Its Own Cloud Accomplishments
Cutting Through the Fog of Cloud Computing Definitions

Photo Credit: NS Newsflash

2 comments:

Unknown said...

Thanks for the clarifications you provide. It really is time for people to bring more attention to the "misinformation" that so many are pushing out these days.

Patrick said...

Thanks for bringing clarity to what companies are delivering when they claim to be a cloud vendor.

Oracle is jumping on the popular cloud bandwagon without really offering a pure cloud solution.