Policy-based data management in ILM.
Business users look to ILM to meet regulatory requirements for data retention and access, or otherwise manage critical business data over its lifetime.
IT hopes that ILM technologies will reduce storage costs and administration by automating the placement, retention, and protection of data using appropriate storage resources.
A comprehensive ILM solution should address both the business user's content-specific goals and IT's underlying data management requirements. Much of the industry's focus has been on content-layer applications, which are some of the first "ILM solutions" to hit the market. But when considering architectures and solutions for ILM, it's important to not lose sight of ILM's potential to reduce managed storage costs by automating data management aligned with that data's value.
Aligning Storage With the Changing Value of Data
Like any other business asset, the value of data changes over time. Infrequently accessed data is typically less valuable to a business than the data used in today's operations. Yet many companies treat both types of data the same by storing them on the same expensive, high-performance disk resources.
Recent data retention regulations modify some data's gradual depreciation by adding regulatory value to the data for fixed time periods. In a regulated environment, data that is older and infrequently accessed has a significant value if required for an audit several years down the road. However, it may not require the performance of high-speed disks. The day after its regulatory period expires, the value of the data may decline dramatically. Some organizations may even decide that the best business decision is to delete the data at that point in time.
By aligning storage cost, capacity, and performance with the changing data value over time, organizations can improve storage efficiency. But manually tracking and moving data is prohibitively expensive and error prone, and can offset any cost advantages of using less expensive storage. What's needed is the ability to automatically locate and protect data according to business values--policy-based data management.
Two Essential Layers of ILM
As a system for managing both the assessment of data value and the actual disposition of digital assets, ILM is comprised of two different technology layers:
* A content-aware layer looks at the content of files and uses that information to approximate the value of the data. Examples include e-mail archiving and image management systems.
* A policy-based data management layer moves and maintains the data on different storage resources based on the storage cost, access, and security characteristics.
These layers work together to create a comprehensive ILM solution. By focusing exclusively on the content layer, you risk losing the data management benefits of ILM.
Dangers of Neglecting the Data Management Layer
ILM is still an evolving market segment. Many early entrants focus primarily on the content layer: classifying and managing e-mail messages, documents, or images. These solutions offer only a rudimentary system for data management, capable of moving objects between a restricted number of dedicated storage resources. For example, an e-mail messaging solution might specify that e-mails more than 60 days old from user Smith must be moved from a given disk volume to a specific tape library.
Although implementing a package like this seems like a quick and easy way to start achieving benefits of ILM, there are long-term drawbacks to this approach. Because the content-layer is application-aware, most organizations will eventually implement multiple ILM solutions across different applications and content areas. If each ILM application has its own dedicated silo of storage, key benefits of ILM are lost.
Buying, deploying and managing distinct islands of storage increases storage hard ware and management costs--something IT is trying to avoid in times of growing data volumes.
When a specific content-layer application manages the storage, it's difficult to switch ILM vendors, add a new storage supplier, use a new computing platform, or take advantage of other technology advances. Companies using dedicated storage risk locking into that supplier for the life of the data.
To reduce costs for data management, storage acquisition (while maintaining flexibility) it's best to create a unified storage environment that can support multiple applications. You'll want to start by creating a policy-based data management environment as a foundation for multiple content-layer applications.
What Makes a Policy-Based Data Management Foundation?
To support a wide range of content-layer applications, the data management layer must provide:
* Unified storage
* Automated data access, protection and retention, based on customized policies
* Support for heterogeneous servers/OS platforms and storage technologies
* A high-performance, salable architecture
Unified storage environment:
The data management layer should create a unified storage environment including RAID, ATA disk and tape. Applications should access the data without knowing exactly where it resides in this environment. Many servers should be able to share access to critical data.
Automated data access, protection, and retention: The data management layer must move, retain, replicate, and delete data according to user-defined policies based on the data's business value. For example, the data management layer should:
* Move data between classes of storage based on policies
* Automatically protect data according to its requirements
* Create multiple copies if necessary
* Manage copies in different locations (including off-site) for disaster recovery
Continuous and transparent access is also vital; the data management layer should retrieve the appropriate data when requested, regardless of its actual location.
Heterogeneous Server and Storage Support: The data management layer must support a variety of operating systems, including Unix, Windows, Linux, and Mac, leaving organizations free to add or change applications as needed. For example, a single data management layer might support an application managing digital images of checks using Linux, another supporting e-mail on Windows, and a third managing customer transaction data logs on Solaris servers. The unified storage environment within the data management solution should be able to include RAID, ATA disk, and tape devices from multiple vendors.
High-Performance, Scalable Architecture: Critical data may require high-speed I/O and large amounts of throughput. In addition, if a single foundation is to create a unified pool of enterprise ILM data, it must be scalable enough to support hundreds of servers. LAN-based file servers cannot generally offer the required performance, as the data must travel over the LAN. In a SAN-based environment, servers read and write data at direct-attached speeds.
ADIC's StorNext software is policy-based data management software that you can put in place today to support content-layer applications while addressing the escalating costs of storage.
StorNext creates and manages a unified storage environment that includes RAID, ATA disk, and tape, moving data automatically and transparently between storage resources based on policies. StorNext ensures that valuable business data is accessible and available at any point in its lifecycle, and that data is protected according to corporate policies. StorNext satisfies user or application requests transparently at any time in the data's lifecycle; the user doesn't need to know where the data is actually stored to access it.
StorNext offers high-performance heterogeneous host access to shared data. Supported environments include Solaris, AIX, Windows, Linux, and Mac, as well as relatively specialized environments such as IRIX (used in many imaging applications) and Unicos (Cray's operating system for supercomputing environments). StorNext creates a storage foundation in which many different content-layer applications can share the same compute and storage resources.
Start by Building the Foundation
If you plan to use ILM, the logical first step is to put the data management foundation in place. With a policy-based data management environment, you can reap near- and long-term advantages:
Immediate return on investment: Policy-based data management starts delivering value immediately, even before you deploy specific content-layer applications. In many cases, users can create simple policies based on file metadata (file age, ownership, last time accessed, etc.) to achieve measurable cost and management savings. For example, you could define policies to automatically move static data sets from the most expensive, active disks to offline storage, reducing costs and speeding backup of the remaining data.
Long-term flexibility: As users prepare to deploy automated content-aware applications, you won't need to redesign the storage environment. A wide range of applications can share the same storage resources, improving storage utilization and simplifying administration.
Like any other emerging technology, ILM presents both risk and opportunity. By focusing only on the content applications, you risk increasing the cost and complexity of storage. Starting with a unified storage foundation capable of managing data according to policies provides the opportunity to improve storage efficiency and simplify storage administration, and ultimately deliver on the promise of ILM.
Mike Palermo is senior product marketing manager with ADIC (Redmond, WA)
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Special ILM Issue; Information Lifecycle Management|
|Publication:||Computer Technology Review|
|Date:||Aug 1, 2004|
|Previous Article:||Unstructured data: the roadblock to effective ILM.|
|Next Article:||Transparent capacity management.|