Although nearly every vendor is eager to talk about ILM, it's often difficult to get a clear picture of what they really mean when they say it.
"Each vendor, customer and analyst, they define ILM in their own different way," explained Raj Choudhary, ILM sales development manager, Asia Pacific of HP. "Why? Because organizations define ILM solutions based on their capabilities and what they have."
He added that ILM implementation is a journey that takes a long time and a lot of resources to complete. The complexity in ILM, according to Gartner, is one of the reasons for the low adoption of ILM among enterprises. A poll conducted by the research firm during its presentation on ILM in December 2004 suggested that only six percent of the respondents have already purchased and implemented ILM plans, as compared to the 36 percent of the participants who replied ILM was not only the planning horizon at all.
Gartner's poll results suggested one of the reasons for a low adoption is that ILM adds complexity by bringing more products, management responsibilities and requirements to the IT shop.
Aiming to address the problem, the Storage Networking Industry Association (SNIA) whose members include major vendors like Cisco, EMC, HP, IBM and Sun Microsystems, formed the ILM Initiative (ILMI) in June. The goal of the ILMI is to unify current efforts around ILM in a unified, cohesive approach.
According to ILMI, "ILM is a new set of management practices based on aligning the business value of information to the most appropriate and cost-effective infrastructure."
The definition suggested ILM is not a software solution that plugs into the existing infrastructure, but rather a combination of processes, practices, policy and technologies.
Hype over reality
However, clearing up this vague definition is not the only way to bring more users' endorsement for ILM. Gartner's report said that another reason for this low adoption was because of ILM's hype over reality.
"ILM can appear to be a daunting amount of work and an 'impossible dream,'" said the Gartner report. "It's also easier not to think about information management and to simply continue to throw hardware and money into storage."
"I think there's still a huge amount of hype," agreed Phil Sargeant, research director, servers and storage of Asia Pacific at Gartner. "Every vendor is preaching an ILM strategy, but I do not think any single vendor has a total ILM offering."
The comment rings true when we look at the SNIA's suggestions on the ILM implementation roadmap. The industry association suggests that ILM be implemented in five phases, starting with data consolidation and data classification followed by matching technologies with data categories, developing islands of automated ILM and, eventually, an enterprise-wide ILM initiative.
Although the SNIA provides a comprehensive roadmap for implementation, most vendors are providing different bits and pieces of the entire ILM roadmap and none currently provide the technology to support the entire roadmap, added Sargeant.
In addition, since [the] SNIA is an industry association formed by vendors, IT executives should "be careful about believing anything from SNIA," noted Jon Toigo, an author and CEO of Toigo Partners, an independent consultancy and technical research and analysis firm. "It is ultimately an industry association-not a consumer group, and most certainly not a standards group."
Find a clear path with data discovery
To cut through the cloud of ILM confusion, both Sargeant and Toigo noted it is essential to first understand your data.
"Understanding the information intimately is the challenge for many organizations," said Sargeant. "I don't think a lot of organizations have understood the information as well as they should."
A variety of information is spread across different business units and applications within the enterprise. The IT team should understand not only who the data belongs to, but also its background, like how users want to use this information and the SLA associated with it, he added.
Although the SNIA suggested data classification early on in the ILM roadmap, Sargeant noted it should really be in the first phase instead of the second.
"I'd say the [SNIA's] first phase is consolidation of data because lots of organizations have already consolidated their data," he said. "For those that haven't consolidated their data, this is putting the cart before the horse a little bit."
He noted it is important to first understand the information before making the next move.
"My advice: understand your data requirements" said Toigo. "Ultimately, I believe that storage [infrastructure] must be tailored to data requirements."
Since data "inherits its DNA from business processes and applications," he said IT should understand data's regulatory conformance requirements, criticality from a business continuity perspective, security and accessibility requirements, retention period, as well as the performance requirements demanded of the hosting platform.
"It is a laborious process that must be undertaken before buying anything from a vendor," added Toigo. "You need to know your data and what it requires from a provisioning and protection services standpoint over time."
This is exactly how the Hong Kong Jockey Club (HKJC) started its storage revamp project early last year. By sending out a questionnaire to users, the IT team researched data requirements from users.
"Previously, [the] service level was inherited," said Raymond Ngai, manager, IT infrastructure at HKJC. "We did not have a clear defined service level and users assumed a certain service level, ignoring the cost associated with that service level."
To help users understand the technical requirements of their data, a relationship manager from the IT shop was also sent to different business units to help translate that business requirement into the technical service level.
"The biggest challenge [in ILM implementation] is classification of data," added Sargeant. "It takes a lot of time to understand the information within the organization because it sits across many stakeholders."
It is not only a very time-consuming exercise, but could also be political, since business units often see their information far more important then the others. "Somebody [must act as] arbitrator to decide who has the most important data," he said.
Seeking outside help, or not
The SNIA suggested that seeking professional services is useful at this stage, and this is what HKJC did. Ngai noted the consultant helped to decide how data should be classified and shared advice on designing the ILM framework for the company.
In addition, a consultant is helpful when it comes to deciding technology approaches, hardware, software, or some of the processes involved, noted Sargeant.
Yet, seeking outside help is not essential, noted Toigo, as long as IT has the proper internal resources and management backing. "It is something that any IT person could do with his or her eyes closed. You just need to collect the information, through interviews and measurements," he said.
Since discovering data requirements is a very organization-centric exercise, Sargeant agreed it should be done internally. But, as organizations move on to start exploring for technologies to apply in its infrastructure, he noted, professional services might be useful.
Step-by-step
As organizations move to apply various technologies over their categories of data, Sargeant suggested to take incremental steps. ILM involves many elements, including backup and recovery, data provisioning, automated data classification and hierarchical storage management.
Since the technology maturity among these elements vary, he recommended implementing ILM by adopting the elements according to their maturity. Elements like backup and recovery are quite mature and very much interoperable in a heterogeneous environment, while the automated classification of data is still very much nascent.
"It's like the old saying, you need to learn to crawl before you could walk," he said. "Organizations have to put in place a lot of different elements that would ultimately form an ILM implementation."
The pillar approach
Apart from prioritizing technology adoption, enterprises should also prioritize the implementation of ILM based on the criticality of data, noted Choudhary from HP.
"Telcos often start with customer data," he said. "But many companies start with email archiving, partly due to the compliance issue."
This is how China Light and Power (CLP) started off its ILM implementation.
"We are using a 'pillar' approach, meaning individual applications," said Andre Blumberg, technology and architecture manager, CLP. "Email and file services are our key priority here and we are actively looking at solutions that exist today for implementation in 1H 2006."
At HKJC, the ILM initiative was applied first on data from its critical applications, said Ngai. The company also implemented ILM on data that needed re-defined service levels.
"After the data discovery exercise, we have a pretty good idea which data needs to [be] replaced within the storage infrastructure," he added.
As an ongoing process to implement ILM across the enterprise, Ngai noted that IT continues to study the data and send out questionnaires to users, as it expands the new ILM storage infrastructure to cover data from other applications and whenever they are planning new releases of applications.
Manual before automation
Under the SNIA's ILM roadmap, automation plays an important part as enterprises move towards phases four and five. Yet very few local users have applied significant automation in the process.
More enterprises are applying ILM strategies with semi-automation, noted Gabriel Leung, general manager, Hong Kong of EMC. For example, container and logistic company Orient Overseas Container Line (OOCL), who recently implemented a ILM strategy with EMC, has only automated the email archiving process so far, said Leung.
"Although many have a defined process concerning their ILM strategy, it still takes time for them to feel comfortable in automating the process," he added.
"Our platform, according to the vendors, can support automation, but none of those automation features were used," noted Ngai from the HKJC. "We still move data across the tier manually."
One major consideration is technology. "Automations are often tools and features built-in in our platforms," said Ngai. "We are confident in auto-provisioning within a single platform, but when it comes to the heterogeneous environment, we are not sure, and we haven't got the time to try yet.
Another consideration is business priority. He explained a lot of planning, like how to create the volume for provisioning and which disks are to create the virtual pool, it all required detailed planning and testing.
"Our team is quite familiar with the process in moving data between tiers, thus automation doesn't really help to do it faster," he said. "Nevertheless, we do have a plan in bringing automation, but we don't have anything concrete yet."
Despite the hesitation among users in applying various technologies associated with ILM, Sargeant noted they are essential eventually, as information is growing so fast that it will become difficult to manage manually.
While vendors continue to provide their own version of ILM and technology may still take time to mature, "ILM is a noble goal," said Sargeant.
"It's nice to give a vision, you like to see the light at the end of tunnel, but you get there by taking incremental steps," he concluded.