Accurate TCO Analysis Demonstrates the True Value of System z
Mainframe evangelist David Rhoderick, shown near his home office on Martha’s Vineyard, Mass., helps give customers a better picture of the true cost of System z. Photography by Tracy Powell
When organizations consider investing in technology, the allure of cheap commodity servers can seem attractive, but it’s shortsighted to focus on sticker price alone without looking at the total cost of ownership (TCO).
However, even when calculating TCO, many organizations fail to find a true apples-to-apples comparison of distributed servers versus the System z* mainframe.
Inaccurate equivalence comparisons, imperfect IT accounting practices, overlooked expenses and other factors can skew the results. To get an accurate accounting, many analysts recommend focusing on the cost per transaction, which gets to the true TCO and puts mainframe investments in a clearer—and price-competitive—light. To learn more about the methodology and challenges of calculating TCO for mainframe, IBM Systems Magazine spoke with David Rhoderick, mainframe evangelist with IBM Software Group’s Competitive Project Office. He and his eight-person team analyze competition for IBM software and hardware, focusing largely on zEnterprise* technologies.
“Accurate TCO measurement and analysis will lead to better decision making that’s focused on BUSINESS GROWTH.”
—David Rhoderick, IBM
Q. From a strategic perspective, what’s the mission of your competitive project group?
A. Our aim is to help customers understand the value of the System z platform and its true cost. We spend a considerable amount of time talking about TCO and investigating how the mainframe compares with other platforms.
Q. What’s the biggest challenge in educating organizations about System z TCO?
A. It’s reaching non-System z people. In the IT world, there are System z proponents (usually those who run the systems) and then there’s everyone else who doesn’t understand it very well—and never the two shall meet, it often seems. A successful approach has been to help our friends in the System z shops educate their colleagues. I’m really talking about bridging the gap to the non-System z constituency and helping them understand various components of TCO.
Q. Once you begin a dialogue with this constituency, is the first step getting them to look beyond the initial investment?
A. Yes, sticker shock is clearly a factor here. Mainframe costs are not very widely known and understood. When you purchase a System z server, you’re investing in an infrastructure that’s designed and configured to grow incrementally with your needs. Once you’ve built the initial configuration, it becomes simple and easy to expand—and also very cost-effective.
Additionally, we find that the System z costs are often overinflated. That’s because the IT accounting process used to “chargeback,” as it’s called, is often imprecise for distributed systems but very precise for the mainframe. Many organizations also include all sorts of unrelated expenses with the mainframe, which obviously makes its perceived cost even larger.
Q. How about an example?
A. We discovered that one customer was including its billing center and call center expenses in the System z chargeback when, in fact, those two organizations are totally independent of whatever platform you’re using. Because the billing center and call center costs were being borne only by the people who paid for the System z platform, users of the distributed systems were getting use of them at no cost. Clearly, this isn’t a good way to run a system or manage the IT accounting of a large corporation. The lesson learned is that customers, in general, need to have a much better handle on what really is a mainframe cost. While the situation in this example was not intentional and evolved over time, it illustrates the issue of accurate chargeback. We’ve even heard of a couple of customers who included the corporate jet in their mainframe costs!
Q. Besides the inexact science of chargebacks, what common mistakes do organizations make when comparing a mainframe to distributed servers?
A. They don’t consider the other associated costs once the system is in place, especially labor costs and all of the other various physical aspects of running IT. For example, there’s networking, electricity, cooling, raised-floor space, firewalls, physical security, printing paper, enterprise license agreements; we could go on and on.