In this sixth tip of a seven-part series, Jon Toigo, CEO and managing principal of Toigo Partners International, and chairman of the Data Management Institute, discusses the hype surrounding cloud storage. Watch the video above or read the text below to find out what he believes are the main cloud storage problems and limitations of the technology.
Please read Toigo's entire video-tip series on data management issues
Data management problems in your data center that need to be avoided
Dealing with server virtualization: Understanding the requirements
Too much hard disk storage in today's data center has a negative impact
Toigo: Don't let price alone determine how you tier your storage
Managing storage infrastructure: Why value-add software doesn't add up
What to consider when formulating a data storage strategy
A real challenge associated with the "storage infrastruggle" (a term I coined to explain the challenges IT administrators face) is maintaining the focus of both IT planners and business decision makers on the problems at hand. Distractions enter the discussion frequently, usually in the form of vendor marketing around "silver bullet" solutions.
Storage clouds are increasingly part of the problem. While billed as the next evolution of storage -- "the great disk drive in the sky" -- there's little that's new or unique about public storage clouds. These service providers are similar in most respects to the storage service providers (SSPs) that were part of the application service provider (ASP) craze of the 1990s or even service bureau computing vendors in the 1980s.
Like their predecessors, contemporary public storage cloud service providers leverage the disgruntlement of consumers regarding the proprietary nature and cost of storage arrays in the market, the management challenges that proprietary gear has created, the friction monopolistic vendor lock-in created, poor vendor performance under warranty and maintenance agreements, and many other factors to make an emotional case for what's essentially storage outsourcing. Not surprisingly, many providers work in concert with a preferred server hypervisor vendor, providing storage offerings specifically for use with that vendor's wares and offering a "less expensive" workaround for the storage disruptions that too often accompany poorly planned server virtualization projects.
Technically speaking, public storage clouds may have developed or acquired some innovative technology to facilitate the delivery of their service to multiple clients. However, on the whole, they have the same technology that's available to anyone building a storage infrastructure. Ideally, they provide storage capacity less expensively by leveraging economies of scale.
That's assuming the service providers reach scale. While numerous (and often vendor-sponsored) trade press articles mine analyst reports to support the idea that adoption rates are high and to the right in public cloud storage, it pays to read the reports from which the citations are taken. In one recent case, an Info-Tech Research Group report was cited in a trade press article claiming that 2012 had seen a 326% increase in cloud adoption since 2010. That sounds impressive, but the revenue increase seemed odd given the preponderance of survey results in the same study that were clearly anti-cloud; for instance, the report cited problems with the cloud like availability, cost, security or suitability to business requirements. For some reason, those survey points weren’t mentioned in the article that cited the adoption increase statistic.
The same study revealed that 88% of "cloud storage" is "storage that is purchased and used behind cloud-based servers." This statistic illuminates one of the problems with the cloud, and with cloud storage in particular: Services themselves are their primary consumers at present, giving a distorted view of the success of the model.
Limitations of public cloud technology include a lack of standards for on-ramps to connect customers to external providers and the APIs needed to integrate services with applications, a lack of management protocols (though, surprisingly, the Storage Networking Industry Association has done some interesting work on a RESTful management approach for public cloud storage that shows promise), and so forth. In addition to a lack of common standards, pricing models vary widely with a growing list of ancillary fees that too often inflate the cost per gigabyte for cloud storage well beyond the miniscule prices vendors advertise. Then there are cloud storage problems like legal issues and regulatory concerns, especially with international operations, and unpredictable service levels.
This same list of limitations and drawbacks existed in the late 1990s and contributed to the demise of SSPs. It's unclear why we should believe that anything has changed today.
Clouds may provide more elbow room for data, and may provide targeted services such as archive or backup that can play a role in certain situations. But these services augment existing infrastructure rather than replace it. Many companies have learned a simple lesson about outsourcing: It works best when the workload outsourced is routine or commonplace. Conversely, outsourcing a problem workload generally doesn't solve the underlying problems, and may worsen them.
When it comes to strategizing about the storage infrastruggle, public cloud services are a distraction, not the solution.