Engates describes the private cloud differentiation this way: "The user is still in the IT business, the software configuration business, the storage management and support business, and the data center business. You still have to have all of those resources."
There is one exception. When a private storage cloud is created by carving off a piece of a public storage cloud for one customer's exclusive use. Users pay a premium for private cloud storage, much like one pays a higher rate for a private room in a hospital.
Essentially, "the difference between public and private cloud storage comes down to the way you connect," says Mike Maxey, director of product management at ParaScale, a Cupertino, Calif.-based vendor whose software was designed to aggregate disk storage on multiple standard Linux servers to create a single, scalable, self-managing private storage cloud.
"If you're connecting over a wide-area network and sharing the resources with other customers, it's a public cloud. This makes sense if you're a highly distributed company and creating applications but don't have a shared infrastructure," Maxey explains. "It's also good if you're putting out transient data, like movie trailers, that might run for five months. Temporary storage in the [public] cloud makes sense."
3. Is cloud storage for all types of data?
No. Cloud storage best handles large volumes of unstructured data and archival material, such as credit card and mortgage applications or medical records. For now, public clouds can't reliably handle highly transactional files or databases that require consistently fast network connections. Any kind of online transaction processing is a no-go.
Cloud storage also isn't an appropriate choice for Tier 1, Tier 2 or block-based data storage, says Jim Ziernick, president and CEO of San Diego-based Nirvanix Inc. "If someone is trying to replace a SAN in supporting a transaction-processing system like CRM, we're not appropriate. Even if we did do block-level storage, the latency of the Internet would cause a noticeable delay," he says.
"What we can do with cloud storage is give users nearly the access that they have with [network-attached storage]," he adds.
Data backup, archiving and disaster recovery are three likely uses for the cloud, says Engates.
"The cloud is for any kind of large-scale storage need with any kind of static-type data," he says. "You don't want to store a database in the cloud, but you might store a historic copy of your database in the cloud instead of storing it on very expensive SAN or NAS technology."
"A good rule of thumb is to consider cloud storage only for latency-tolerant applications," says Terri McClure, a storage analyst at Enterprise Strategy Group in Milford, Mass. "Backup, archive and bulk file data would all do well in the cloud, where subsecond response time is not a requirement." On the other hand, databases and any other data that is "performance-sensitive" aren't suited for cloud storage because of latency, she notes.
But before moving any data to a cloud, public or private, users need to address a more fundamental question, says Mark Tonsetic, program manager at The Corporate Executive Board's Infrastructure Executive Council.
"If you go to cloud storage, does it solve the problem of understanding where and why storage growth is out of control and where the point of value is [in storing a particular set of data] in the entire end-to-end business process? Just moving the technology to a cloud is not an optimal solution," Tonsetic says.