Will the promise of ILM remain elusive?

Given the plethora of storage TLAs (Three Letter Acronyms), such as SRM, HSM, ILM, DLM and CAS, and new storage mantras from appliance-ize ways to gateways to cram it in your rack and pray ways, storage managers deserve lots of praise--not to mention lots of Valium. And while most new storage management technologies hold the promise of lower TCO and golf scores in the nebulous future, and lay claim to the Holy Grail of Information Lifecycle Management (ILM), alas, there remains a dearth of new technologies that can actually deliver the proverbial chalice.

According to a recent GIS-TICS study, $8,200 per person, per year, is spent on file management activities for search, discovery, classification, prioritization, indexing and management of critical files. In fact, recent surveys conducted by BusinessLaunch at the 2004 Compliance Seminar Series clearly indicate that this is fast becoming one of IT's highest priorities. Why?

Enterprise Strategy Group (ESG) reports that regulatory compliance is driving a 174% annual increase in disk-based storage growth. After November 15, 2004, all public companies over $75M market cap must now comply with the Sarbanes-Oxley Act (July 15, 2005 if under $75M). Failure to comply could mean severe fines and imprisonment for corporate executives. Furthermore, fines and imprisonment can accompany the suppression or destruction of evidence. The legal term for this is spoliation, which can occur if the proper reporting mechanisms and data security measures are not in place. According to a recent survey of 350 public companies performed by A.R.C. Morgan, over 60% of CFOs resign or are "pushed" when a significant material weakness is disclosed. And more than 86% of these material weaknesses are discovered by the audit firm. By then, it's too late.

Ignorance of the law, as always, is no excuse. CIOs must now be 100% certain that their IT infrastructure and procedures comply with the COBIT framework standards for Sarbanes-Oxley 404 regulations. The internal ability to identify and analyze the operational data for potential transaction integrity or control weaknesses must be in place, as well as technologies that can discover, classify, prioritize, manage and report on all types of information. Fines for non-compliance can easily approach $750,000 or more.

This is why business Service Level Agreements (SLAs) are escalating--not only in quantity, but also in complexity. At the 2004 Compliance Seminar Series, one large insurance company CIO grumbled about an SLA that required a seventy-two minute mandatory recovery for all critical files. His problem? Classifying what's critical to his business users. Compound that by the fact that classifications and priorities can change weekly, if not more often. How does one deal with complex, ever-changing environments where tens of millions of files remain unclassified? Where the promise of ILM remains ever elusive? Where antacid and prayer remain the only means of coping?

ESG and other analysts agree that unstructured datafiles now account for over 80% of corporate data. It's also a well-known fact that current file systems (CIFS for Windows and NFS for UNIX) typically provide no more than eight basic metadata summaries (file name, extension, date, etc.). Also, current relational databases, designed for structured transactions, are inflexible, expensive and not suited to the management of unstructured data (files).

Limited by the file system and tied to relational databases, today's ILM and related solutions, including SRM, HSM, ECM (Enterprise Content Management), DLM (Data Lifecycle Management), e-mail archiving, data migration, CAS (Content Addressed Storage), tiered storage, etc. are all debilitated by the limitations inherent in files systems and databases. They are therefore incapable of Information Lifecycle Management as defined by SNIA (Storage Networking Industry Association), specifically, "ILM is comprised of the policies, processes, practices, and tools used to align the business value of information with the most appropriate and cost-effective IT infrastructure from the time information is conceived through its final disposition."

The key phrase above is "align the business value of information ..." How does one do that with only eight colors in the metadata crayon box? How does one do that with solutions tied to relational databases that are inflexible to business value changes--if you didn't plan for it in advance, you probably can't do it. Certainly not without a DBA and a data model rebuild. So now what?

Enter Information Value Management. SNIA documentation, industry analysts and experts agree that the first step in ILM is data classification. The question of "what is it?" and, most importantly, "what is its value?" must be answered before "where does it go?" In short, we need to identify the individual value of what is in our closets, at the file level, then clean and organize that mess before we can prescribe a means to keep it neat and tidy for the future. One solution that claims to do this is EMC Documentum's Content Storage Services (CSS). Neville Letserich, director of product marketing at EMC, stated that, "Organizations need to be able to classify content, and this requires a level of detailed information about data that other storage management applications can't provide." ESG's Marrone-Hurley noted that "Documentum provides more than static attributes and that's where they differ [from some other products]." CSS's policy engine assigns specific tags to the files based on preset attributes and then places the content on the appropriate storage device and automates the process of moving it from tier to tier as needed. Auditing and migration logs allow for IT tracking and allocations as well as rich reporting. Apparently, CSS solves the ILM dilemma.

But wait, there's a catch: CSS only works in a Documentum environment, which requires a large investment in dollars and integration time. Also, CSS's entry price tag exceeds $50,000. And you still need a relational database, which is inflexible. To further salt the wound, Documentum typically requires a file format change to manage data. However, it is an effective and robust solution for those shops with a healthy budget.

Abrevity (a start-up founded by Joel Harrison, co-founder of Quantum Corporation) claims to have solved the problem from the ground up. According to Abrevity, its FileBASE Information Value Management software empowers granular file discovery & analysis, dynamic classification & prioritization, and automated file & e-mail management. Using a combination of meta-data extensibility and customization, combined with a flexible and dynamic "free-attribute" database, this software promises to deliver the industry's first non-intrusive, automated and query-based IVM policy engine. If true, this would let professionals automate the process of discovering, classifying and prioritizing file selection and management for archive, backup, disaster recovery, tiered storage, charge-backs, compliance and legal audits and other ILM requirements. Recently demonstrated at the ILM Summit in Irvine, CA, this software boasts independence from ECM applications, such as Documentum, and compatibility with virtually any hardware or other ILM software products.

"FileBASE is the first ILM solution to leverage a purpose-built non-relational database that allows for dynamic flexibility, scalability and compatibility," according to Harrison. "Our dynamic schema allows for fast, easy adaptation to customer requirements, which is difficult or impossible for solutions tied to fixed schema databases."

Abrevity's core technology, developed over the past two years, requires no IT or user disruption and has been tested extensively by customers. These customers report that file name words, directory words, creators, dates, types, sizes, etc., contain rich metadata information that when properly parsed, queried and customized, can provide virtually all of the parameters needed for their IVM requirements. Taking a different approach from in-document search-based software, FileBASE is designed to "narrow the universe" of files via metadata classification, and then search inside only the remaining uncategorized documents. It does this by first extending metadata from the few basic attributes given by the file system to hundreds of divided, combined and customized attributes. These attributes can then be queried to locate and retrieve files for IVM reporting and management. Finally, it allows custom "tagging" of queried files for future retrieval, as well as query-based ILM file management (replication, migration, deletion, etc.).

"Experts agree that the first, and most critical step in ILM is granular, detailed data classification," says Harrison. "The problem is, ILM vendors don't offer this in their solutions. As a result, IT professionals haven't been able to relate ILM policies to business Service Level Agreements."

Bill Reed is vice president of BusinessLaunch, Inc. (Livermore, CA)

www.businesslaunch.net