MYTH 3 :

Low software costs come from good vendor discounts.

Reality: A discount is relative to some cost structure based on a commonly accepted set of assumptions around hardware configuration, software configuration and pricing metrics. Alter the metrics and you dramatically alter the baseline cost the vendor uses to determine a discount.

In other words, just because a discount looks good on paper does not necessarily mean it is good, since discounts can be and sometimes are, above list price based on actual usage.

And the discount you get often depends on the region you are in. "A typical data centre will spend $5500 per MIP on its software costs. Negotiated discounts represent between 5 and 15 percent of the total savings opportunities," Swanson says.

"Product inventory is the single largest contributing factor to software costs. Over 35 percent of mainframe software products have replacements available. A best-in-class company might pay a greater amount for its software even though it typically utilizes 36 percent fewer products than an average data centre. The reality is that over 80 percent of savings opportunities are directly attributed to managing product inventory and configuration and have nothing to do with discounts," he says.

MYTH 4 :

Low software costs come from having a large number of enterprise licence agreements (ELAs) with best-in-class terms and conditions.

Reality: The theory goes that many enterprise licence agreements are built around best-in-class terms and conditions. Negotiate numerous multi-year deals with really good contract language, and you can expect to enjoy low software costs.

In practice, though, many such deals damage data centres by locking them into a commitment that may have been good for the business environment of the time but that is totally unsuited to any new business reality. "Businesses change so rapidly today that if you make a three-year commitment, your business is going to be so different in three years time than it is today that you will find yourself committed and locked into something that you shouldn't be," Swanson says.

When data centres were growing rapidly in the 1980s and 1990s there is no doubt ELAs helped, he says, but in slower growth years much of the cost is soaked up in unused inventory and capacity. ELAs often feature a "waste factor" of more than 50 percent to pay for unessential products and capacity.

"A common problem associated with ELAs negotiated prior to 2001 was that data centres were growing at 20 to 35 percent annually and thus wanted large capacity amounts imbedded in their ELAs," he says. "Since 2001, data centre growth has slowed dramatically, leaving many companies with excess capacity. The notion that best-in-class companies negotiate large ELAs as a cost protection was true during high growth years. As old ELAs were renewed, they were negotiated with similar product configurations."

Part of the problem is that companies do not allow enough time to review their inventory mix and configuration in order to understand the current value of their data centre's core products. Organizations thus sign long-term commitments based on a set of assumptions built on the current business environment, which often features many inefficiencies in the way the data centre is run. Commit to a long-term deal and you commit to the inefficiencies.

"ELAs are [also] padded with bundled-in products and discounts, in much the same way that a 20-course meal might be offered to someone on a diet. The reality is that best-in-class companies have the fewest ELAs and are smart in buying only what they need when they need it," Swanson says.