Complete ILM solution doesn’t exist today, but that doesn’t mean you shouldn’t seek the benefits of data classification.
By now, most IT executives have heard of the importance of having an information life-cycle management strategy. Not surprisingly, much of the ILM buzz has been generated by storage vendors eager to jump on the ILM bandwagon.
“If you sell dog food today, you’ve got an ILM story. Even people with the most obscure product have an ILM story,” says Richard Scannell, vice president of consulting for GlassHouse Technologies. “The real issue is that you can’t buy an ILM solution today. There’s no ILM in a box.”
Most industry analysts agree with Scannell that ILM is not a product suite you can buy from any one vendor. Nor will it be in the near future.
“One thing I note when I work with customers is that ILM and how to get there is nothing that’s a big bang theory. It doesn’t happen immediately or is something you can buy off the shelf,” says John Herrera, vice president of global solutions operations for EMC , an early proponent of ILM.
But that doesn’t mean ILM should be ignored either. If you define ILM as a set of practices or processes that lets IT treat data based on its usefulness or value to the business over the life span of the data, then few IT managers can deny the potential value of ILM.
At its core, ILM involves classifying data based on varying degrees of importance to the business (such as mission-critical vs. operational).
Once data is classified appropriately based on its value, ILM strategies involve developing rules, service-level objectives (SLO) and management policies that address how each class of data should be treated. These rules define how the data will be stored, protected, migrated and accessed over time to or from different tiers of storage (such as primary, secondary and offline).
ILM and the data center
So, what’s the benefit of ILM? Michael Peterson, program director for the Storage Networking Industry Association (SNIA) Data Management Forum – which is spearheading its own ILM initiative and five-phase ILM road map – says: “The vision we have for ILM is as a new management process for the data center. The No. 1 problem in the data center today is complexity. The No. 1 problem with owning and operating storage is complexity. . . . The real power of ILM is to reduce the cost of operations by 100x – by solving the complexity problem.”
At the final phase of the forum’s ILM road map, which Peterson estimates at anywhere from five to eight years away, there will be automated, end-to-end ILM solutions that run independently of any one business-side application, can scale across a company’s geographic and organizational boundaries, and will automatically store, protect, replicate, move, copy or archive data to various storage media based on how a company’s own policies and SLOs have defined the data over time.
The hope of ILM is that data management software will evolve to automate the process of managing such policies and the underlying data. Peterson says he expects to see a new breed of ILM management utilities and storage operations management software entering the market as a means to address this issue.
Even at this early stage, Peterson says there is still immediate benefit to IT managers who place themselves at the start of the forum’s road map. Citing a number of cases where companies have begun to follow the road map, he says: “They are classifying and setting service levels and already finding that going to that level has begun to improve their operations. People are starting to get benefits of beginning to work this way.”
EMC, Veritas on the minds of enterprise IT
Taking at least the first few steps to develop an ILM strategy appears to be gaining traction among enterprise IT managers. Preliminary findings from a recent study of more than 100 Fortune 1000 companies, conducted by research firm TheInfoPro, show ILM as one of the top five storage management technologies IT plans to implement in 2004 or 2005 (see graphic, bottom of page).
Two-thirds of study participants indicated they had some type of ILM strategy in place now or that they were planning to implement an ILM strategy over the next few years.
According to Ken Male, founder of TheInfoPro, some of the early themes around ILM in the study include growth in tiered storage, disk-to-disk and Serial Advanced Technology Attachment, and growth in use of other storage hardware media such as network-address storage products and mid-tier storage array offerings. “Our customers are telling us that they are tiering their storage – but it’s really setting them up for a future ILM strategy,” Male says.
When asked to indicate which vendors they planned to use over the next few years to assist them with implementing ILM, study participants placed two storage vendors in the lead: EMC, followed by its rival within the data management software market, Veritas Software .
EMC came out of the gate early to articulate the benefits of ILM and educate IT organizations on EMC’s own ILM vision and three-step implementation framework (see graphic). EMC’s recent spate of software-related acquisitions, including Legato Systems, Documentum and VMware also have been carefully positioned to bolster EMC’s ILM vision.
| ILM by the numbers | |||||||||||||||||||||||||
| |||||||||||||||||||||||||
It may look relatively easy, but achieving full ILM could be a five- to seven-year process.
Deploy tiered
infrastructure
Deploy application-specific ILM
Enterprise-wide ILM
Not to be outdone, Veritas has acquired data management-related software companies. (Recent acquisitions include Precise/WQuinn, Invio Software and U.K. e-mail archiving vendor KVS.) According to Jeff Lundberg, senior manager of Veritas product marketing, “Veritas really espouses the utility computing vision – which incorporates a lot of the components conventionally lumped in with ILM. Realistically, we’re saying data life-cycle management and information life-cycle management are part of what these IT organizations need to do – but on top of that is the need to do other things as well, such as reporting, charge-back and application performance management.”
Enterprise Strategy Group analyst Nancy Marrone-Hurley connects ILM to utility computing this way: “Information life-cycle management is a subset of a utility environment, also known as an on-demand, adaptive or autonomic environment. [With ILM], there are significant cost savings [IT managers] will realize in capital, in hardware cost savings as it relates to storage, in administrative cost savings and, more importantly, in being able to better protect their data. In the long-term, if anybody ever wants to implement a utility environment, they need to implement ILM processes first in order to lay the groundwork. You can’t expect the underlying infrastructure to be adaptable if you haven’t set these processes up.”
Many say that once this initial groundwork for ILM is laid, choosing the software or hardware technologies to support ILM becomes a much easier proposition.
“There are a couple of things IT organizations need to understand about ILM: It’s 90% process, 10% technology,” says Phil Goodwin, a former senior program director for Meta Group, now with Diogenes Analytical Laboratory.
His advice to IT managers embarking on the first ILM steps? Spend your first few months learning how data is being utilized in the organization. Focus attention on the pain points surrounding your critical applications and data. Expect to spend 12 to 18 months on critical applications. Then, start over with your less-critical applications, he says.
Develop no more than three to five classifications, such as mission-critical data, business-critical data and operational data. “All of these imply the information is important to the organization. You can’t afford to lose it,” he says.
Then develop rules to identify more-specific data characteristics. Goodwin says most organizations would benefit from the establishment of an IT or storage policy director to spearhead this effort. “This policy director, for lack of a better term, is someone who is responsible for establishing these rules and interfacing with the business units,” he says.
From there, he suggests handing off policies to system architects who can develop the appropriate solutions and infrastructures to support them.
While ILM does not happen overnight, EMC’s Howard Elias, an executive vice president in the company’s office of technology, agrees with Peterson that customers can start seeing early fruits from their labor.
For example, Elias says consultants might look at customer data stored on high-end Symmetrix arrays and be able to identify 25% or more that he’d classify as “inactively accessed fixed content.” In these cases, organizations which move that data immediately to a lower-cost alternative, such as the company’s mid-tier Clariion offering, will immediately see better use from their storage resources.
The second step in EMC’s ILM road map suggests IT look to specific applications – such as e-mail systems and company databases – as key candidates to help automate predefined policies and migrate data classes to archives. This is where products such as EMC’s EmailXtender, DiskXtender and DatabaseXtender come in, he says, to help deliver policy-based data movement across storage tiers.
EMC’s final ILM implementation step is to use a common workflow-based policy engine that tracks and moves data across multiple applications across the corporation, Elias says. “Even though [the data] came in from SAP or an e-mail database, it will have the same policy engine. This is where it truly becomes enterprise-wide, application agnostic,” he says.
EMC sees its Documentum product fitting this bill, although Elias acknowledges the company and the industry still have a way to go before they can help IT organizations get to this step.
