Within the new data center, data moves from one storage resource to the next based on information life-cycle management policies.
| |||||||||||||||||
Chris Smith, senior manager of computer services for semiconductor maker PMC-Sierra, knows he can save money by making storage decisions based on the value and business criticality of the information stored in his data center. Working with business managers to determine importance, he stores the frequently accessed, most crucial data on expensive primary storage while relegating less important, infrequently accessed data to inexpensive secondary and tertiary storage. Like many, he is slowly gaining an understanding of a new form of data management that comes part and parcel with the move to the new data center.
Smith calls this new data management scheme “tiered storage,” but it also is known as information life-cycle management (ILM) or data life-cycle management.
Whatever the name, this technique addresses how a company manages data from the point of its creation to its disposal. ILM consists of a number of methodologies, such as problem assessment, socialization – working with business partners to determine data’s value – classification and review, says Nancy Marrone-Hurley, senior analyst for the Enterprise Storage Group (ESG).
“ILM is a process that enables you to move data down a path of storage resources, from a high-performance, high-capacity filer [or disk array] to a lower-end disk array, to a tape silo so it’s near-line, and then finally to a permanent archival media,” Smith says.
At PMC-Sierra, Smith reserves Tier 1 storage for the production data used most frequently. This type of data – information on semiconductor products still in use – accounts for most of the company’s storage. Ninety percent of PMC-Sierra’s stored data belongs in Tier 1, he says.
In Tier 2, he uses inexpensive Advanced Technology Attachment (ATA) drives to store data that isn’t as fresh or needed as often. In Tier 3, he uses tape to archive data that is accessed even less frequently.
For ILM to work within the new data center, companies need to consolidate their storage. They can do so by using virtualization or via an integrated core-to-edge director-class switch if they have storage-area networks.
Within the new data center, server capacity and storage resources are virtualized, data moves from one storage resource to the next based on a company’s ILM policies, and storage and provisioning tools reallocate storage dynamically. Server and storage become inseparable (note EMC’s acquisition of VMware for its server virtualization software, and Veritas’ purchase of server automation vendor Jareva and application performance management company Precise Software Solutions). As these technologies take root, the result is far more flexible, efficient and cost-effective storage environments.
Jim Doedtman, technical planning manager for OSF Healthcare, a nonprofit healthcare corporation in Peoria, Ill., recently finished classifying data as part of an ILM program begun last year. Through this process, the firm found it could move about 70% of its transactional data from high-end EMC Symmetrix and Clariion Fibre Channel arrays to EMC’s lower-cost Centera arrays. Fixed content data – medical images, patient registration forms, information that never changes – will reside on the Centera storage. “The other 30% is data we are keeping online based on frequency of use and age,” Doedtman says.
“We have EMC Symmetrix as Tier 1, Clariion Fibre Channel as Tier 2, Clariion ATA as Tier 3 and we’ve moved to EMC Centera for Tier 4,” he says. While tape is traditionally the last tier, he says OSF Healthcare has created disk-based storage tiers because retrieving data from disk is faster and easier compared with getting data from tape.
But today’s technology only can take ILM users so far, ESG’s Marrone-Hurley says. Smith and Doedtman, for example, must move data from one tier to another manually. What they need is the ability to set policies that trigger the automatic movement of data based on its content, creation date or last access.
| The life cycle of data Like hierarchical storage management before it, information life-cycle management uses a tiered approach. But HSM moves data based on age alone while ILM determines tier based on business value. Click here for more |
Such automation is the aim of start-ups such as Arkivio, KVS, MessageOne and OuterBay Technologies, as well as the reason EMC acquired Documentum in October 2003. EMC plans to improve the intelligence of its tiered storage capabilities through the integration of content management software it gains from that vendor.
Multivendor interoperability and integration of disparate storage management applications also come up short today.
“Users with multiple storage management applications have to figure out how to integrate applications that support ILM,” says John Webster, senior analyst for the Data Mobility Group. “I’m not sure the vendor community has stepped up to that yet. The proposition that seems to be out there is ‘buy everything from us and we’ll integrate it.’ The user isn’t going to stand for that.”
Marrone-Hurley cautions that enterprise storage managers shouldn’t expect the ability to invoke integrated end-to-end, policy-based ILM for a few years. “There are a number of products that can do certain aspects of ILM, but interoperability and levels of integration among software and hardware are not here,” she says.
Analysts and vendors say the Storage Networking Industry Association’s Storage Management Interface Specification will promote interoperability and integration of ILM applications. Users await that work.
“Most of the ILM software is fairly new. For our environment, we don’t have a good fit yet,” Smith says. “I am quite interested to see how things go in the next 12 months and who percolates to the top in providing a solution that really works.”
