Just how realistic is the vision of storage as a utility service? Is it a distant dream or a near-term reality? And most importantly, how do we get there from here?
Storage and other parts of the IT landscape may well be trending towards a model similar to that offered by the utility industry. In other words, picture a scenario where you pay your monthly bill for services used and don't have to pay any attention to the internal plumbing. If you want to alter your service, you just make a call, tell an operator your request, and they proceed to set you up with a different kind of service plan.
But just how realistic is this vision? Is it a distant dream or a near term reality? And most importantly, how do we get there from here? This article looks at the utility concept for storage and offers some answers.
Industry analysts seem to agree that there is a definite trend towards storage and network management functions becoming far more closely integrated and better automated. This is an essential building block on the road to utility-like storage.
“This reflects industry revolution towards integrating new technologies and domains with mainstream enterprise management,” says Dennis Drogseth, an analyst with Enterprise Management Associates. “Vendors are looking to find more effective ways to normalize and organize data, as well as to broaden access to multiple types of management applications through partnerships, standards, acquisitions, and other approaches.”
Certainly, a utility-like model would be desirable — seamless service availability for the end user without the need to have to deal with the nuts and bolts of everyday storage management. Much like attaching any type of electrical device to the mains — you just plug it in and it works. On the other hand, that analogy is probably too simplistic; there are, after all, many more variables in storage and IT than exist in electricity.
A better example, then, might be telephony. Today, phone service is a relatively simple utility, despite the immense systems and networking complexity running behind the scenes. You can choose your options, such as local and long distance rates, digital, analog, mobile or not, etc., but the basic infrastructure is simply there for the user, and it (usually) works.
If you want more rollover minutes or unlimited long distance, you simply place a call, a telephone operator makes a few clicks, and hey, presto!, your service has been changed. Neither you nor the phone operator had to do anything difficult, because a large team of behind-the-scenes telecom staff has automated everything to the point where the complexity is completely masked.
Page 2: Getting There from Here
Getting There from Here
In the same way, you may soon be able to tailor your storage costs by choosing service levels of redundancy, availability, latency, throughput, application feature sets, and so on. As an example, perhaps you have a budget of $x, so you consult with a storage utility rep to see what type of storage plan will get you the most bang for your buck. Before such a utility model develops in storage, however, standards will have to become better accepted in the industry.
Let's use e-mail as an example of how standards eliminate plumbing complexity. Ten years back, competing email systems hung together poorly and interoperability challenges were prevalent.
Today, regardless of the mail software you are using, whatever hardware platform you are on, and from wherever you are connecting to your mail server, you just send and receive. Email now universally runs on the same simple standards — SMTP, IMAP, and POP. Standards transformed e-mail into the “killer app” of e-business because it is universally demanded and seamlessly delivered to the user.
Similarly, network technology has standardized to the point that it is rarely visible to anyone outside of IT. Networks took about 30 years to get up to the sockets level whereby the routers and switches are now cloaked in invisibility.
“Storage doesn't have that long to evolve,” states Mark Bradley, Storage Architect at Computer Associates (CA). “It has to quickly move out of the proprietary software band and into standards-based interoperability.”
Industry standards would appear to be most immediately relevant in the areas of discovery (finding out what is there) and enumeration (specifying exactly what, where, and in what quantity it is). By standardizing these areas, we would arrive at the basic protocol requirements to build on.
Fortunately, over the last six months we have seen the development of industry-standard interfaces for SAN discovery and enumeration of Fibre Channel fabric at the API level. That is gradually being expanded to cover NAS, iSCSI, bus-based storage, etc. Once accomplished, the next level will be complete control and command at the API level.
“We have the same basic stack as networking but many variables like Ethernet, fabric, and SCSI,” says Bradley. “We must integrate these so we can manage across all platforms to encompass proprietary software point products for storage management, backup, SAN management, and so on.”
The good news is that the various vendors are actively working to build the bridging pieces between their various applications so they can be viewed, if desired, on one screen. The ultimate goal is be able to associate a certain storage service to a specified QoS or an application, and have it automatically delivered — a function some vendors are calling dynamic provisioning.
Products are already arriving on the market with some basic dynamic provisioning capabilities. And within twelve months we will probably see a high degree of dynamic provisioning evolve between the various storage management systems vendors.
However, don't get too carried away just yet. CA's Bradley estimates that it could take a further 10 to 15 years before we have a truly common storage environment that duplicates the utility industry in terms of seamlessness and simplicity.
This feature originally appeared on Enterprise IT Planet.
See All Articles by Drew Robb