Although end users are convinced of the benefits of information life-cycle management, they are delaying adoption until better tools become available.
By Michele Hope
Love it or hate it, the term “information life-cycle management” (ILM) is here to stay. Results from a recent survey of Fortune 1000 storage professionals conducted by TheInfoPro research firm (www.theinfopro.net) show that 62% of the respondents have an ILM implementation either currently underway or slated to begin later this year or next year.
When such findings are compared to TheInfoPro’s survey conducted six months earlier, however, a rockier picture of ILM emerges: Users appear to be delaying their ILM implementation plans.
For example, in the earlier survey 32% of the companies planned ILM deployments in the “near-term,” or sometime in 2005. However, in the more recent survey that number dropped to 17%, with 36% now placing ILM implementation into their long-term plans for 2006 (see figure).
According to Ken Male, CEO and founder of TheInfoPro, survey respondents attribute the delay in ILM implementations to a number of technological and procedural issues, including the inability to automatically move data between storage tiers and the lack of robust chargeback mechanisms that allow IT to pass some of the cost savings of lower-tier storage back to data owners in the organization.
While most of the respondents appear sold on the benefits of ILM, many still await the maturation of ILM-related tools as they struggle to move past the concept stage into actual ILM deployments, says Male.
According to one survey respondent, “ILM has moved out to the long-term plan. We are hoping for it to be the mechanism for automating movement of data between tiers. I know that is not the full grandiose vision of ILM, but it is a tactical need. At this stage ILM is still a concept.”
Overcoming the ‘Big Three’
Stephen Foskett, director of strategy services at GlassHouse Technologies, sees three technical obstacles to more-rapid ILM adoption: a lack of metrics, lack of granularity, and lack of data movement tools.
“If I put a record in a database, or a file in a file system, ILM has to have metrics on it. ILM has to know something about it besides what it knows today,” says Foskett. “ILM also needs to handle granular pieces of data, like a record in a database, instead of the whole database. The third requirement is automatic movement of data. All of these are big technical challenges.”
Even without such tools widely available, many respondents to TheInfoPro survey are forging ahead with deployments that fit under a larger ILM umbrella. These include the use of tiered storage with Serial ATA (SATA) disk arrays, midrange arrays, and NAS devices for backup, reference, or archive data that does not require the highest level of access, performance, or uptime attainable from more-expensive storage (see figure below).
Despite the fact that enterprise-wide, end-to-end ILM and related data management tools are not yet available, and standards work by organizations such as the Storage Networking Industry Association (SNIA) is still in the early stages, some ILM proponents are still upbeat. For example, Michael Peterson, program director for the SNIA Data Management Forum and a spokesperson for the forum’s ILM Initiative and road map (see figure below), points out that IT organizations that have approached ILM in the right way have been able to demonstrate a return on investment (ROI) in as few as six months.
According to Peterson, the keys to initial ILM success involve starting with smaller pilot projects, keeping data classification (see table below) and policy development relatively simple, and focusing more on ILM “solution stacks” already available to help manage an organization’s unstructured and semi-structured data.
This usually involves integrated solution stacks currently targeted at file-based or e-mail-based archiving or enterprise content management (ECM). Start-up Kazeon is one example of a company that hopes to enable initial ILM implementations with an appliance that searches and helps classify unstructured data and associated metadata (see “Kazeon manages unstructured data,” InfoStor, April 2005, p. 12). Other emerging players in the data classification and policy-based management arena include vendors such as Abrevity, Arkivio, Scentric, and StoredIQ (formerly Deepfile).
Peterson says that vendors making notable progress in the development of integrated, end-to-end ILM solution stacks include EMC, IBM, Network Appliance, and StorageTek. “EMC is ahead of the rest, but vendors such as IBM and Network Appliance are coming along very strongly as well.”
“Right now, there are a few key ILM components that can be delivered in a transparent, non-disruptive way that are focused on the archival of information to speed performance, limit risk, and reduce costs and complexity. These include e-mail, file system, and database archival, as well as business continuity, data protection, and ECM,” says Brett Cooper, senior manager of ILM and compliance at Network Appliance. “ILM cannot be accomplished in one step. It is a strategy that today encompasses point solutions for specific needs that will in time be connected by larger management solutions and technologies.”
Of the types of applications most suited to initial ILM deployments, e-mail is often the “beach head,” according to Gary Lyng, director of marketing, ILM products and solutions, in Hewlett-Packard’s StorageWorks division. Lyng cites “e-mail control for both TCO reduction and compliance” as key areas where early adopters are making progress with ILM. Other areas include file movement that incorporates primary storage, management, and archiving, and database archiving to reduce TCO, improve overall performance, and meet compliance requirements, says Lyng.
The need for education
SNIA’s Peterson says that the bumps in the road to ILM are part of the natural adoption cycle in any early market and that user education will be the key to the future success of ILM in IT organizations. “You don’t have to be at phase four [of the SNIA ILM road map] where everything is fully implemented and automated in order to say you have an ILM environment,” says Peterson. “A company doing simple archiving and tiered storage is doing ILM. The easiest way to proceed is in phases, where you put in place a backup tier or a compliance store, even though you may be doing the data transfers manually.”
The University of Texas at El Paso (UTEP) is one organization that has ventured into ILM and is currently creating storage tiers to handle backup, disaster recovery, and a new e-mail archiving project. The university uses a mix of storage systems from EMC, including more than 5TB of production data housed on a Symmetrix DMX800 and a Dell/EMC CX600 subsystem.
According to systems administrator and IT manager Luis Hernandez, the university went “live” this month with automated archiving of Microsoft Exchange 2003 e-mail via CommVault’s QiNetix DataMigrator software. E-mail archives will be available to any of UTEP’s 25,000+ students, staff, and faculty and will be migrated onto a DMX800-based SAN whenever the e-mail passes a certain age or size. Hernandez and his team have configured the system to allow users to voluntarily “opt in” to the archiving program to allow the creation of an “infinite in-box.”
GlassHouse’s Foskett says that his firm often advises clients to minimize the risk of an ILM deployment by only deploying a new, lower tier of storage for new storage requirements, as opposed to attempting to migrate existing data. Foskett also recommends incorporating internal storage tiers within a single high-end array.
Foskett also offers the following advice when you’re researching ILM solutions: Spend more time investigating the type of support you’re likely to get from prospective vendors. “People spend too little time looking at vendors’ support organizations and too much time looking at features they won’t use. Can the vendor respond in two hours if you have a problem?”
A sampling of vendors shows that most have begun to put significantly more emphasis on building support organizations and services that address ILM. For example, HP last month rolled out a slew of services under its StorageWorks ILM Services Framework. StorageTek has also refocused its support organization to better support ILM.
Other large vendors such as EMC and IBM also offer a wide range of ILM-specific professional services that encompass the assessment, planning, and implementation stages of ILM deployments. Network Appliance offers its own set of professional services, as well as combined services in conjunction with key partners. Hitachi Data Systems’ ILM services include storage economics strategy services and application-optimized storage assessment and planning services to aid in data classification and building efficient infrastructures that reduce complexity.
Although applications such as e-mail archiving appear to offer a fairly straight-forward initial foray into ILM, experts maintain that a firmer understanding of what ILM is and isn’t is critical to an organization making further progress.
The SNIA DMF defines ILM as a “standards-based, business-driven management practice,” with an emphasis on the word practice. “ILM is not a service; it’s a practice,” explains Peterson. “ILM isn’t a management tool; it’s a management practice. That’s a key difference.”
EMC defines ILM as “aligning IT with the needs of the business by maximizing the value of information at the lowest total cost over the life cycle of the information,” according to Dennis Hoffman, a vice president of marketing at EMC. “Simply moving data from one tier to another tier lowers total costs, but it doesn’t necessarily do anything to maximize the value of information.”
“We urge users to look at their environment, measure it, and pick small opportunities to develop some ILM momentum,” advises Todd Rief, senior director for remote services worldwide at StorageTek. “Once you’ve done it in e-mail, for example, you can roll it out to your database environment, and [add applications] until you get to the end stage-policy-driven ILM across the enterprise.”
“ILM is really about people, process, and tools,” says Jim Geis, director of storage solutions at Forsythe Technology. “If you can separate those three elements, you can then look for the tools you need to automate the process.”
Geis and his team recommend an ILM framework built first from a larger corporate, legal, or compliance policy that is usually established by others outside of IT, such as a compliance officer or legal department. One rule of thumb during the policy creation phase is to apply a “CARD” filter to each piece of company data, which evaluates data based on questions surrounding its Creation, Access, Retention, and Deletion, according to Geis.
Over the next 12 to 18 months, a number of technologies are expected to emerge that will make it easier to move to an ILM framework. Storage virtualization is one of them, according to William Hurley, a senior analyst with the Enterprise Strategy Group consulting firm. Hurley sees virtualization as a key enabler that will help IT organizations manage information assets from a more logical viewpoint. Virtualization technologies allow the storage “container” for application and server data to be managed as more of a logical representation than as a specific physical box. Virtualization also allows physical storage to be more loosely coupled with server and application components.
According to Hurley, fabric-based virtualization players to watch include IBM (with its SAN Volume Controller, or SVC), Cisco, and EMC (see “EMC ships virtualization platform,” p. 1), as well as smaller vendors such as StoreAge and Troika. Host-based virtualization players he notes include Veritas with its Volume Manager, Clustered File Server, and data management and backup products that allow data movement to be handled from within a single software platform. PolyServe is also a player to track, according to Hurley.
“Another class of technology addresses the use of metadata,” says Hurley, citing EMC’s Centera and continuous data protection (CDP) technology from vendors such as Network Appliance, Revivio, and TimeSpring.
DLM first, then ILM
Hurley makes the distinction between data life-cycle management (which he believes most users are doing now) and true information life-cycle management.
“Users going down the ILM road are solving data life-cycle management problems first, such as backup,” says Hurley. “They are beginning to think more about the data itself: Is it new, old, critical? They are then using features such as volume copy, snapshots, CDP, etc., to begin moving data to less-expensive resources to get a bigger bang for the buck.”
Proponents believe that any IT organization stands to benefit from embarking on the ILM journey. “It’s a way to change how IT organizations run and manage their information technology-from applications to disk,” says EMC’s Hoffman. “This is what we’re focusing our company on.”
Michele Hope is a freelance writer. She can be reached at firstname.lastname@example.org.