Servers also require administration and management–important components of indirect server cost that should be considered when purchasing a new system because of the complexity of today’s IT environments. In most cases, corporate data centers rarely manage just one operating environment.
Rather,
companies typically use a combination of Windows, Linux, and UNIX operating systems, as well as an occasional proprietary product. These mixed-OS environments may be the result of mergers or acquisitions, but they usually occur naturally as IT organizations and business units select the best applications to fit their business needs, regardless of the operating system.
The problem with mixed environments–Mixed environments, however, create costly administration and management challenges for IT organizations. Because each of these operating environments has traditionally required specialized hardware to support it, IT departments are often forced to utilize diverse types of servers that may or may not come from the same hardware vendor. For example, customers who run UNIX and Windows typically need two infrastructures because each operating system requires a different platform.
Each separate platform requires the IT department to build a separate support organization around it.
For each platform, there are costs associated with training requirements, the constant administration and support needs, the efforts to expand or grow the hardware environment, and the investment in the necessary support infrastructure. These costs are magnified as organizations add new operating systems and required hardware to their IT environment.
IT organizations caught in a costly administration-and-support cycle for a heterogeneous environment often assume that they have no choice but to invest in the support organization required to manage each environment. The fact is, IT organizations cannot simply ignore or limit the support of a hardware environment that the business considers important; in some cases, a hardware failure can cost a business thousands of dollars per minute.
Other IT organizations are choosing to streamline IT operations by forcing the business to standardize on one operating system and hardware platform. This approach enables IT staff to be trained on a single type of hardware, follow one purchasing approach, have one set of spare components, and call one support number–resulting in reduced administration and support cost and increased responsiveness to the business. However, it limits the business’s ability to select the best applications to fit their business needs, regardless of the operating system.
Ideally, IT departments need a single platform that can run multiple operating systems; this approach would streamline IT operations because there would only be one platform to support, but it would still provide the business with the flexibility to choose the best applications to meet its needs, regardless of operating system.
Comparing server capabilities to streamline IT operations–For discussion purposes, we will compare the options customers of HP, IBM, and Sun have available to them if they need to support, for example, both UNIX and Windows (discussed in the prior example). In this case:
• An IBM customer would need to purchase both the p690 to support AIX and the xSeries system to support Windows.
• A Sun customer would need to purchase an SF 15K to support Solaris and, because Sun servers do not support Windows, another server (from a different vendor) to support Windows.
• In contrast, an HP customer would only need to purchase the HP Integrity Superdome–a single platform–to support both operating systems.
Note that only the HP Integrity Superdome provides a single, common hardware platform that can run UNIX, Windows, or Linux–and even run them simultaneously on one partitioned server. As a result, the Superdome provides a unique solution to the high cost and management headaches normally associated with supporting mixed environments. The Superdome reduces administrative and management costs via a single platform approach; increases business agility by allowing the business to run multiple operating systems and maximize server utilization via partitioning; and streamlines operations by eliminating the need to create multiple support organizations for multiple servers running diverse operating systems.
Regardless of the operating system(s) an organization is running–UNIX, Windows, or Linux (or all three)–its IT staff only need to be trained on the HP Integrity Superdome. The department only needs to stock supplies and parts for one hardware version. When problems arise, the IT staff has only one number to call for help. As the business grows and changes, IT staff can simply install additional cards and set up a new partition for a new application or test environment–regardless of the operating system–rather than bring in a whole new server for these purposes. And finally, the Superdome can be easily upgraded and redeployed as applications on one operating system are retired and new applications requiring different operating systems are added.
Upgrade costs (long-term investment protection)
In the months and years after an IT department acquires a new server, vendors typically make available new processors and functionality that can be purchased at an additional cost. In order to take advantage of these new features, IT departments must engage in an upgrade process. For customers evaluating servers to purchase, it’s important to understand the types of upgrades each server enables–because all upgrades are not equal in terms of cost and required effort. Some upgrades can be completed with a simple in-box upgrade (i.e., a cell board swap in the case of Superdome); other upgrades may require an entire box swap.
The cost of upgrades is frequently overlooked in initial TCO analyses conducted prior to purchase decisions, primarily because customers assume that all servers have the same upgrade features (regardless of vendor)–and that taking advantage of these upgrades will cost the same. In reality, there are often huge differences between the types (and costs) of upgrades from vendor to vendor.
And these differences can dramatically impact the long-term investment protection provided by–as well as the TCO of–the server.
Comparing box-swap versus in-box upgrades
In general, box-swap upgrades cost more to implement than in-box upgrades, primarily because it takes significantly longer to uninstall a server from the network and then install a new server (during a box swap) than to simply pull the cell boards out of the chassis and put in new ones (during an in-box upgrade).
More specifically, with a box-swap upgrade, costs are higher due to the following:
• Higher personnel costs. As explained above, it takes significantly longer to de-install a system and install a new server. In addition, IT staff must reload the operating system and applications, train personnel, test the environment, etc.
• Longer downtimes. Studies indicate that downtime of high-end servers running mission-critical applications can cost companies anywhere from hundreds of thousands or even millions of dollars per hour. Even for planned downtimes, the costs can be substantial. As a result, maximizing server availability is paramount. Because a box swap requires the server to be taken out of the network, it causes significantly longer downtime than a simple in-box upgrade–and, as a result, it costs more to implement.
• Opportunity costs. By engaging skilled personnel in a lengthy box-swap upgrade process, the IT department loses the opportunity to have those personnel complete other priority projects, such as the deployment of a new marketing application that boosts revenue generation. In addition, because box-swap upgrades take longer to complete, the time-to-deployment for a new application is extended, which in turn delays the return on investment.
• Depreciation. A box swap frequently mandates a change to the system’s serial number. Given the tax laws, this can potentially impact the company’s financial depreciation schedule and expenses.
In addition, changing the serial number often has a direct impact on support contracts and software licenses, requiring either a transfer cost or new license fee. Even if direct costs like these can be avoided, the process of obtaining the new serial number takes time and costs money, further increasing the total cost of the upgrade.
When one combines the cost of the above items and then factors in the actual hardware costs, the price difference between a box-swap upgrade and an in-box upgrade can be staggering.
