Information tends to change its value to a company over time. Information lifecycle management (ILM) seeks to allocate information to various storage resources according to its current value. Thus, as a data set becomes less significant it can be transferred from a company's best storage assets to less costly storage.
An example of this might be the sales department's January sales figures. These are important in January and are likely to be used to compare against February's numbers, but by autumn, the figures are only of historical interest. As their value shifts over time the data can be migrated from expensive Fibre Channel arrays to less expensive primary storage, and then transferred to whatever levels of storage are available until finally they are archived. In a well-maintained data lifecycle environment all this would happen automatically, according to prescribed management polices.
I have mentioned in the past how important a role ILM is likely to play in determining this year's storage buying habits. HP rolled out the latest version of its ILM strategy last week at its two-day analyst conference in Massachusetts, and there is no reason why I can't use this column to share some of what HP told us.
HP's recent acquisition of Persist Technologies has provided the company with a foundation product on which to base its ILM strategy. The solution, called AppStor, can be thought of as a RAID device that understands content.
The AppStor product is built to address a major problem that affects all ILM approaches (and a lot of data warehousing as well). Most existing methods of data retrieval must face the fact that, as the total mass of data gets larger, the act of searching for and retrieving a particular piece of information from that data will slow down proportionately. Unfortunately, this direct relationship between archival growth and a slow down in the act of data retrieval conflicts with a key facet of ILM: the need to harvest information efficiently and quickly.
AppStor attacks the problem with an inside-the-box grid-like architecture that adds processing power to the array as the data store gets bigger. HP claims this enables it to guarantee answers to queries in less than three seconds, no matter how big the archive becomes.
HP will focus its ILM solution at the health care, financial services and life sciences industries - as do the other ILM vendors. Those sectors, after all, are the ones that have to cope with the scary regulations that are causing CIOs the biggest nightmares.
I suggest that most of us should be circumspect. When it comes to ILM, we should understand that no product, however good it may be, is ever likely to provide a completely packaged ILM solution. Technology is important, but it is only so within the context of an appropriate set of supporting processes. These processes will help companies understand the value of data throughout its life, and will provide the underpinning for all those policies that eventually will automate the way that data is managed.
At this point, if we think of ILM as being 10% technology and 90% process we may not be too far wrong. For certain, we should keep an eye on those vendors that may have the 10%, but let's make sure that as we do so we also pay attention to developing our own set of supporting processes. It is a good bet that only by doing so can we make sure that any investment in this technology be a success.
This last point opens up the whole issue of the soon-to-be-lucrative business of ILM consulting... but we'll dine on that can of worms some other day.