Easier Storage Management: Are We There Yet?
SMI-S, Aperi, StorageRevolution.com, HP/AppIQ it seems there's no shortage of efforts to simplify the world of storage. SMI-S has been in development for years and enjoys massive backing from a veritable Who's Who of the vendor community under the umbrella of the Storage Networking Industry Association (SNIA).
More recently, however, IBM appeared to break ranks with SNIA and formed its own open standards body known as Aperi along with Sun, Brocade, Cisco, Computer Associates, Engenio, Fujitsu, McData and Network Appliance. Jon William Toigo then created a stir with StorageRevolution.com, but this area has been relatively quiet since the initial announcement. Meanwhile, HP is developing its vision for an HP-based storage management platform combining elements from its various storage software solutions to technology acquired from AppIQ.
So when is all this going to result in a life of leisure for the storage manager or administrator? Mike Karp, an analyst with Enterprise Management Associates, said he believes that SMI-S is the only significant standardization effort currently in existence. He says SMI-S is headed down the right path. Rather than trying to solve everything in one fell swoop, it does a single task well device discovery. But for now, the benefits are somewhat limited from the end-user perspective.
"The major value of SMI-S comes to the vendors right now as it reduces their development costs," Karp says. "But it has made things easier for storage administrators in terms of discovery."
SMI-S focuses on Host Bus Adapters (HBA), storage array and switches. Karp, however, points out that the zone of discovery is widening as tape vendors and other OEMs obtain SMI-S certification.
"As more of them become certified, it's going to be even easier to do device discovery," says Karp.
In simple terms, SMI-S is the standard management interface that allows different types of storage hardware and software products supplied by multiple vendors to interoperate for the purpose of monitoring and controlling resources.
"Today, SMI-S can help users ease configuration, discovery, provisioning, trending, event management, security and asset management tasks," says Rob Callaghan, chair of SNIA's Storage Management Forum and senior product manager at ADIC. "Additionally, it increases customer choice of solutions, reduces the number of management packages required, increases the ability of the consumer to perform asset management and provisioning of heterogeneous storage resources, reduces agent proliferation across the enterprise as well as reduces overhead and complexity of managing ever larger amounts of storage IT or storage manager."
SMI-S will eventually replace all of the proprietary or common management activities that users do with various storage devices, Callaghan say. However, the level of innovation within the industry is causing the standards body some problems. Callaghan says that it takes time to create a standardized implementation for each new piece of functionality or technology and to add it into SMI-S. Thus SMI-S development can't go much faster than a 12-month release cycle.
"Vendors are only able to deploy the technology as it lines up with there existing product lifecycles and these frequently do not always align," he says. "Also, customers are not always able to upgrade every time a new release is deployed as they must first validate the release, budget for it and many times thoroughly test it before it can be deployed in a production environment."
How does this relate to the various Aperi and the other movements and platforms mentioned above? Callaghan says SMI-S is a standard that can be used and leveraged by initiatives like Aperi and others. For example, for HP and AppIQ, SMI-S is the underlying language used to communicate with their devices.
Aperi versus SMI-S
What's the difference between Aperi and SMI-S? McData is in the interesting position of being heavily involved in both initiatives. The company's director of SAN solutions, Tom Clark, explains that creating open standards isn't the same as creating an open management platform.
"You can write proprietary management platforms that make standards-based API calls to solicit device information or perform configuration changes," he says. "Aperi's goal is to have an open systems management framework based on standardized (SMI-S) objects."
The idea is to put both the standards and the management framework into the public realm, as opposed to having open standards but a number of proprietary management frameworks. He admits that, at least to some extent, the formation of Aperi at a point in time when SMI-S is still in a 1.0 standard is an effort to move more quickly from the requirements definition/formulation phase into working product. That's why none of the vendors who signed up for Aperi have quit SNIA's SMI-S effort.
Aperi, Clark says, pushes for an open code management framework that would leverage SMI-S. But that doesn't mean SMI-S is a less important element. He touts the advanced storage features of the upcoming SMI-S version 1.1 such as storage virtualization.
"I don't think there are any inherent limitations on what SMI-S could help manage," Clark says. "There needs to be various forms of advanced intelligence in the storage network to actually execute higher-level function such as automation, storage policy execution, proactive traffic shaping, etc. but theoretically any of those high-level functions could be controlled by standards-based APIs."
The first 15 years of SAN technology, Clark says, have been devoted to getting the infrastructure into place and working. The next 15 years, he predicts, will focus on advanced storage services, with SMI-S becoming a common language for coordinating services such as policy-based storage administration, information lifecycle management, continuous data protection and aligning the storage infrastructure to more fully support the needs of application data.
What about HP? Clark characterizes AppIQ's Storage Essentials (SE) as a good example of a for-profit proprietary management application that uses standards-based APIs (SMI-S): "I suspect this is what other storage vendors would also like to do, especially ones who have not joined Aperi."
HP remains firmly committed to SMI-S while rolling out its own HP Storage Essentials platform, which makes heavy use of the old AppIQ StorageAuthority Suite. It's said to deliver heterogeneous SAN management, storage resource management (SRM), and provisioning for NAS, SAN and direct attached storage on a common platform.
"From a single management console, administrators can manage their complete HP and heterogeneous server and storage environment with a feature rich, extensible and secure management tool set," says John Kelly, product manager for HP's StorageWorks division. "The unified platform offers shared server and storage core services such as auto discovery, inventory management, distributed tasks, event notification, reporting, central repository, GUI/CLUI, role-based security and Web service API's."
Taking a different tack, StorageRevolution.com views the efforts of SNIA and SMI-S as too vendor-centric and not really catering to the needs of end users hence the lack of traction in realizing the dream of storage manageability. Like Aperi, it's an effort to create a open systems management framework.
"StorageRevolution.com is being formed purposely outside the storage industry and vendor community, so it would have to rely exclusively on individual contributors," Clark says. "Without the expertise of those who have actually created the storage network infrastructure, a third-party effort is going to take a very long time."
Two Steps Forward Three Steps Back?
Another way of looking at the storage management burden, though, is that we may well have made huge strides in recent years in terms of manageability. Unfortunately, that progress has been obscured by another factor the explosive growth of storage environments and the fact that storage gear can now sprawl across corporate networks.
Think back a couple of years back and storage gear was relatively centralized. If that environment still existed, today's management tools and standards would have made major inroads in lessening the load on the storage administrator. Unfortunately, the gains made in streamlining of management have probably been outstripped by the demand to have bigger, better and more complex storage infrastructures that span regions and even continents.
"Everyone wants bigger storage but not more complex storage," says Karp. "The problem is that it appears to be very hard to have one without the other."