* Price of widgets are down but what about the cost of managing those widgets?
The price of the highest performance (15K rpm) SCSI drives is now below $6.50 per gigabyte. On the other end of the scale, the price for low-end SATA drives has dropped to about one-tenth this amount. Quantity won. I am guessing most of the readers of this column buy their storage in somewhat larger quantities, and thus their price per gigabyte is going to be substantially lower.
The price of the highest performance (15K rpm) SCSI drives is now below $6.50 per gigabyte. On the other end of the scale, the price for low-end SATA drives has dropped to about one-tenth this amount. Quantity one. I am guessing most of the readers of this column buy their storage in somewhat larger quantities, and thus their price per gigabyte is going to be substantially lower.
As my friends on the software side of the business are fond of pointing out, “storage is free, it’s the management that costs.”
Well, maybe. The dollars used in buying all that storage hardware come out of somebody’s budget – I rarely come across a hardware vendor who really is just giving the storage hardware away or an IT manager who tells me that his “money bucket” is infinitely deep. But the point is well made. Those operational costs associated with managing the storage infrastructure just keep on coming, quarter after quarter after quarter. It’s the operational expenses that gets you.
Most IT managers try to understand the total cost of ownership (TCO) of whatever they purchase, but if truth be told, most fail miserably at the job. It is just too hard for them to come to grips with all the associated expenditures that may appear as “trailing costs.” Price of course is always important, but what about all the things that may contribute to cost during the lifecycle of the devices you buy? I wonder how many of my readers are consistently willing to purchase products that are not the cheapest choices available?
Simplistic TCO analyses have always been a blight when it comes to contributing to IT decision-making. After all, they make the fundamental assumption that cheapest is best – but if cheapest is always best, why don’t Fortune 500 companies trust critical corporate data to cheap devices? Because they know the cost of managing many cheap devices is often exorbitant. More importantly of course, they also appreciate the potential breadth of costs associated with data downtime. So they choose something as bullet proof as they can find, swallowing the up-front cost in favor of long-term, deferred savings and a good night’s sleep.
“Cheapest” can only be determined on the day a piece of hardware is retired from the IT room, written down by the accounting department, and donated to charity.
“What this country needs is a good five cent cigar,” said Thomas Riley Marshall, Vice-President of the U.S. in 1919, probably a somewhat simplistic analysis of how things were back then.
Several decades later Franklin P. Adams added the following rejoinder: “There are plenty of good five-cent cigars in the country. The problem is they cost a quarter. What this country needs is a good five-cent nickel.”
All of which I take to indicate that the real cost of things – cigars, storage arrays, whatever -is much more than just the purchase price. It also means looking at whatever else may contribute to cost throughput the product’s life. With cigars, this meant accounting for inflation. With hardware, above all else it means accounting for the costs associated with manageability and with downtime.
Economists and business people have understood this for a long time. Technologists have not.
