"Headroom," when used in the context of storage, represents that extra mass storage we keep on hand for when our users unexpectedly feel an urge to store a few extra terabytes of data.
Managers often try to protect their sites against wanton or unauthorized use of storage resources by assigning quotas and the like, but we also have to provide for responsible users whose needs are caused by legitimate data growth. And so we buy more storage. Most IT managers have some sort of rule of thumb that tells them to purchase an extra 10% or 15% more disk space than they think they will really need. Just in case. The result is that headroom grows incrementally with each quarter's purchases.
Because so little of our storage is really managed completely, this typically results in a self-perpetuating cycle of escalating quarterly expenditures and increasing amounts of unmanaged disk space. The disk drive companies love this, of course. Our chief financial officers do not.
It is more apparent than ever that enterprises need to become increasingly agile if they are to survive. They must provide for what is needed now, and also for what will be required tomorrow. If the goal of IT is to supply a completely reliable and optimized set of assets and services in support of their company's needs, managers need vendors whose products will deploy applications quickly and scale resources and services without impacting applications. This means nothing less than supporting the continuous transformation of the IT infrastructure itself.
Providing for this kind of environment is a fundamental concept behind the current movement towards utility computing, which aims to provide users with hardware, software and services on an "as-needed" basis. Whether utility computing is implemented through an internal IT group or by way of an external service provider, the model is the same: users will access (and in theory, pay for) only those assets they use. What happens to the storage that isn't being used? It goes back into the general pool of assets, and remains available for the next user whose needs exceed his current means.
When discussing utility models, this extra, unused pool of assets - idle disk space when it comes to storage, but also unused ports on a switch, unexploited bandwidth, and so forth - is sometimes termed "white space." White space is what most of us have been referring to as "headroom" for the last five years or so, but with one major difference - when the utility model works properly, assets are deployed and redeployed on-the-fly rather than through "quarterly provisioning" via more and more purchasing.
Properly managed, storage assets can be reprovisioned whenever the need arises. If you have begun to suspect that virtualized storage is a key enabling technology in support of the utility storage model, go to the head of the class.
We'll discuss the implications of this next time.