Despite the fact that some parts of the cloud are loosely defined, not everybody is bringing products to market worthy of the name "cloud." As a matter of fact, there's a lot of cloud washing--renaming existing products, after a few tweaks, with the word cloud inserted.
John Cullinane, former chairman of Cullinane Software, started the trend many years ago, when he took a pretty good database product, IDMS, and in the midst of an emerging, relational database technology trend, renamed it IDMS-R. "There," he said. "We're relational too."
To actually be part of the cloud requires systems that run on a simplified data center architecture, operated largely by automated policies, not human hands. The architecture allows end users to self-provision their own servers, and has a billing mechanism that allows the supplier to charge only for the resource used, not the lifetime software license cost.
[ What's Salesforce.com's grand cloud plan? See Benioff's Dream: Toppling Microsoft, Oracle, And SAP. ]
The cloud environment also extends to the end user's ability to start, stop, or modify his own applications. That makes it different from the previous generation's outsourcing services. The provider, in turn, can scale the environment by simply adding servers, network bandwidth, and disks that become part of the pooled resource. And remember, the cloud is supposed to be resilient. Hardware devices may fail underneath it, but with the right architecture, the cloud workload keeps running.
NIST has supplied the only reliable cloud definition, although I say it amounts to more of a description of existing cloud deployments than a precise definition. That means that "cloud computing" is a new way of doing things that hasn't shaken down into a strict definition--it's still a moving target. At the same time, it's only fair to say that some vendors are exploiting that opening and cloud washing their products.
On Wednesday, Dec. 14, cloud consulting company Appirio will host the Cloudwashies—an "awards" event similar to the Razzies awards for bad movies. I've got my own list of nominees for top cloud washers--those parties who have done the most to try to make a previous generation product look and sound like cloud.
1. Oracle Exalogic Elastic Cloud
At Oracle, you couldn't find someone willing to use the word "cloud" in a complete sentence before Oracle OpenWorld in September, 2010, when Larry Ellison did an about-face and finally started using the term himself. After that, the needle on the chutzpah meter went off the scale and one Oracle product after another not only had cloud attributes but also was suddenly "the cloud." Until then, Ellison had been denouncing cloud computing--and even in the turnaround speech, he offered instruction on how to detect a cloud washer at work.
So my number one candidate for cloud washing is the Oracle Exalogic Elastic Cloud, a name that contains so many contradictions of the definition of cloud computing that it threatens to render the term meaningless. It's an old fashioned appliance that's been renamed "a cloud in a box," when we thought the cloud couldn't be put in a box. Granted, some automated administration applies, thanks to all that highly engineered integration, but where is the end user self provisioning, charging based on use, those economies of scale that are supposed to be part of the cloud environment, and the escape from lifetime licenses? What's in the box, judging by the price tag, is a whole bunch of lifetime licenses for previous generation software. This is pre-cloud middleware and applications wrapped up in a cloud bow.
2. Lawson Cloud Services
The warning flags should have been unfurled when Lawson Software senior VP Jeff Comport took to the YouTube airwaves in April, 2010, to announce that Lawson had full cloud services, unlike those vendors relying on multi-tenant architecture, such as Salesforce.com. Whenever you hear a putdown of the capabilities of Amazon.com, Microsoft, Rackspace, Terremark, or Salesforce.com, cloud washing is sure to follow.
What Comport was saying was that Lawson's financial, budgeting and other core applications had been made available in the Amazon cloud As Lawson Cloud Services. They had not been reduced to lowest common denominator applications services, as Salesforce.com's were, he claimed: "In order to pursue multi-tenancy, they've had to commoditize the application, dumb it down…"
This was similar to Larry Ellison's attempt to attack Salesforce.com for offering "insecure" multi-tenant database services (even though Salesforce.com uses the Oracle database.) It is now established that the boundaries set by the machine in a multi-tenant environment are sufficiently hard to protect the privacy of data and business processes of users, as they use the same server and software.
But cloud washers, lacking other cloud features, trumpet on-premise applications made available in an Amazon data center as the real cloud thing. It doesn't work that way. If anything, it's much more difficult to architect cloud applications that take full advantage of multi-tenancy and pass along to customers the economies of scale that follow.
3. Storage Vendors Who Don't Come Clean
Storage was in many ways the first cloud service and major cloud providers made storage a part of their service portfolio at an early date. Hence, I was surprised to see two years of tests of 16 top cloud storage services offering a range of variable reliability. The two years of tests were conducted by Nasuni, a supplier of storage-as-a-service through various cloud vendors. Amazon Web Services' S3 service was ranked as most reliable.
According to Nasuni,, Amazon's S3 was available 100% of the time on average each month, during its two year test period. In fact, S3 experienced 1.43 outages a month, said Nasuni, but they were of such short duration that they were unlikely to be noticed by a customer.
Likewise, three other companies were tied at 99.9% availability, Microsoft Azure, Rackspace, and Peer 1 Hosting, with Nirvanix nearly in the same league at 99.8% availability, according to Nasuni. Number six was AT&T Synaptic storage-as-a-service at 99.5%. AT&T had few outages per month "but their duration impacted its availability," the study reported.
Nasuni tested 16 services but only disclosed the results on the top six. The gap is big enough (100% to 99.5%) between the top six to make me wonder what the bottom six look like.
In other words, some storage services listing themselves as using the cloud approach have not yet mastered something that should be taken for granted in the cloud, the reliability factor. Part of the cloud computing mystique is that cloud services are resilient. Hardware devices fail, but the services continues running without loss of data.
If we took 99.5% reliability as the minimum to be expected of a cloud storage service (that's 3.75 hours of downtime in January,) what should we call the 10 vendors below the minimum? And who are they, I asked Nasuni. "We don't want to disclose that information," was the response. Well, Nasuni leaves a lot of legitimate participants in cloud storage that it may not have tested as suspects. How about full disclosure, next time, to protect the innocent?
4. HP Cloud System
Former HP CEO Leo Apotheker showed a certain proclivity for commissioning grand cloud initiatives, while behind the scenes, HP was busy rolling up previous generations of software into some things with cloud names. On March 15, Apotheker described an Internet service integration platform for the 100 million WebOS devices that HP would sell over the coming year. The mobile platform was an integral part of his strategic vision for an "HP Cloud."
Only a few weeks later, the first HP WebOS device, the TouchPad tablet, was launched, then soon abandoned. Other WebOS devices failed to make it to the launch pad. WebOS has been recently jettisoned as open source code donated to the community, and Apotheker moved on to new challenges, outside of HP at the board's request, as it replaced him with Meg Whitman.
His platform in the sky as a strategic vision came back down to earth as simpler and plainer cloud services, wrapped up what's become known at various stages in 2011 as the HP Cloud System. In fact, many HP cloud operations depend on pieces of its existing system management software.
Cloud Service Automation, part of Cloud System, does include gleanings from the 2010 Stratavia acquisition, which tracks configuration changes in deployed systems, useful in launching cloud workloads. But much of Cloud Service Automation is a reassembly of predecessor products, such as HP Network Management Center, the former OpenView, and HP Performance Center. Together, they provide configuration management, monitoring, and deployment in the cloud, as they did earlier for physical assets in the data center. HP's services consultants say they want to do more Cloud Discovery Workshops for prospects. Maybe a place to begin would be HP's own marketing department, so that it learns to blur the line less between old and new.
5. Oracle Public Cloud
Yes, Oracle deserves a second entry in my list. Oracle Public Cloud is a service that uses cloud in its name in such a way that tends to render the term meaningless. (See number 1.) The public cloud does not consist of highly engineered appliances that run vendor specific systems, on which only one vendor's software may be installed.
Anyone tried to run Red Hat JBoss or IBM WebSphere middleware on the Oracle Public Cloud recently? It's as if Mr. Ellison came across a whole warehouse of unsold Exadata and ExalLogic machines and said, "What's this? I've been telling Wall Street we sold all we could build. Get rid of 'em." The result: Oracle Public Cloud, built out of Oracle appliances.
As a development platform, I get it--that part makes sense--and software as a service consisting of Oracle applications and middleware. That's fine, also. Just don't call it public cloud. In real "public" cloud computing, choice of and control over what's running has passed to the end user.
Charles Babcock is an editor-at-large for InformationWeek.