"Go through your applications to find out what you have. You then need to understand the capacity of each of those things and what the growth forecast is for the next three to four years. That may take some time, but it is time well spent as this information will set you up for later negotiations."

McIsaac then creates a spreadsheet of the data picture and assigns labels based on importance: Platinum for high-value data that needs to be available 24/7 with no data loss and back within four hours; Gold; Silver (for things like archive); and even Bronze if necessary.

"Group the data into a small number of classes -- that is classes of data, not classes of disk."

And this is a key point. The classes are based on the data, not the technology. From here, you can sum up all the capacity and growth rates. The result is you have a picture that says you need X amount of terabytes in Platinum, X amount in Gold and X amount for Silver year by year. Write this up as a storage strategy document.

Now you understand your data you can go to the vendors with a tender.