More Is More

The most significant development in the storage tier of ILM is the "flattening" of the infrastructure required to manage the information to the extent that the distinction between operational and warehouse storage environments has become blurred. For the ABC this means the elimination of manual tape handling, dramatically reduced information retrieval times and the ability to make automatic replication copies as its content ages.

"The warehouse is an integral part of ILM anyway," Knowles says. "It sits in the background and comes forward when you need it. If you look at a lot of corporate data, you would keep it there whenever you had space and eventually you would pension it off on usage, and so forth."

Short, having also conducted research on warehouses, says the data supports the idea that warehouses have grown to the point that in many companies there is little distinction between warehouses and operational level data, driven in part by the declining cost of storage and the fact that it is easier to buy another storage rack than it is to go down the path of an enterprise data management strategy.

Consequently all organizations are seeing an explosion of data storage because "it's easy and cheap", Knowles says, adding that storage capacity as a means for ILM is equivalent to building a shed at the back of your house - you keep throwing things into it but eventually you will have to do a clean-up. "You think it doesn't cost you much to store it but it takes you a few days to clean the shed," he says. The ABC is now distinguishing between discretionary and non-discretionary data. It will manage the former as a business obligation, and charge users for the latter with the rationale that people will not store data unnecessarily if they have to pay for it.

How to store information efficiently is the fundamental challenge of storage-tier ILM, because "we can't forever keep adding more and more", Knowles says. Even in the context of the ABC's video and audio material, it could go to the extreme of keeping every camera tape ever taken but much of it may be redundant, even in the long term. Conversely, Knowles urges CIOs to be aware of how the value of information may change over time.

"Our natural history people say we do take a lot of footage but if the animal disappears it may become very valuable," he says. "A lot of the value of information comes from re-purposing it. Corporate data has value because of the other things you do with it. It's looking for trends long term and looking at data a totally different way. If I look at the data from a different angle, does it tell me something different?"

Knowles says part of designing an information model is about trying to segment storage in a way that delivers "what makes sense" if the organization wants to look at it in different ways. "What does it cost me to produce Radio National? I can't tell you that if I don't collect that for a particular service," he says. As an example, the ABC recently completed an exercise to better classify its assets. It had records of how old the assets were, but did not know precisely what they were being used for. Variables like which programs made use of the assets at what periods of time, how they all related to each other and the utilization factor of most of the material remained a mystery.

"We are now looking at that from different perspectives and saying, let's look at it in content streams, let's look at it from platform streams," Knowles says. "Then we can make decisions on how important it is for us to replace the asset, or is it time for us to reappraise the way we deliver?"

Short agrees, saying an equally important variable to consider in ILM policy is the type of information rather than just the application or the business function that created it. "Defining information types is more important in dealing with unstructured rather than structured data," he says. "For the latter, if I know the application and the business unit, I have most of what I need. In unstructured data, I really need more, and a schema that fits strategy to information types makes a lot of sense."