The most complex challenge for today's CIOs is to increase the information technology (IT) organization's ability to rapidly respond to new business needs while continuing to cut costs. Reducing fixed/non-discretionary spending is key to increasing IT agility; and an often overlooked tactic is optimizing the IT hardware refresh cycle. A company's IT hardware refresh policy drives decisions on millions of dollars in annual capital spending and operating expense. An effective refresh policy will guide IT and accounting personnel through the analyses needed to balance the trade-offs between capital spending, operating efficiency, and risk mitigation.
However, in many companies, the IT refresh decisions don't strike this balance because they are driven by just one parameter -accounting depreciation schedules. Standard 3 and 5-year refresh periods are applied across all IT hardware, leading to unnecessary spending in some areas and unplanned risk of failure in others. This paper will present a systematic approach to optimizing an IT hardware refresh policy -- unhooking technology decisions from accounting's useful life calendar - in a way that drives savings in capital, labor, and operating costs.
An IT hardware refresh policy is a formal or informal standard that guides the replacement or update of hardware. A typical policy consists of two components: the type of hardware and the number of years the hardware is to be used before refreshing it. The stated purpose of most hardware refresh policies is to avoid putting a company at operational risk by retaining older, less reliable, and/or unsupported hardware.
In practice, both formal and informal hardware refresh policies are tied to depreciation cycles. The effect is that finance and accounting groups set refresh cycles rather than the IT organization. For example, if a business sets a 3-year useful life for a piece of hardware, the typical policy will call for that hardware to be refreshed in 3 years. While accounting standards are an important consideration in the development of a refresh policy, they should not be the main consideration. Businesses across industries -and even within the same industry - have significant differences in how they run their operations and in the applications that support them. These business and technology requirements should be the main driver of the hardware refresh policy.
Unoptimized refresh policies keep businesses from capturing the full value of their refresh spend. A refresh policy that is too loose can miss out on operational cost savings. Many companies have informal refresh policies with a de facto guideline of, say, 4 years. However, because the refresh policy is just an informal guideline, business groups don't budget for refresh, using the existing hardware until it breaks down or the supported application is retired. These companies may have saved capital dollars, but often at the cost of higher operating costs in higher power consumption, higher hardware maintenance, and higher labor costs.
Storage hardware provides a good example. Data storage is an area of explosive growth for many data centers, tempting many CIOs to ignore their refresh policies and keep operating existing storage hardware for as long as possible. However, operating costs for storage hardware can sharply increase over time. Many storage vendors drastically increase the price of hardware maintenance in the fourth and fifth years of ownership and new units can be significantly more power efficient on a raw per-gigabyte basis. A properly constructed hardware refresh policy will guide you to make the right trade-offs between conserving capital and missing out on lower maintenance and power costs.
On the other hand, when the policy is too tight, technology dollars are wasted. For example, a company has a formal 3-year server refresh policy. Every year, a third of its servers are replaced with new servers regardless of the criticality of the application, the technology roadmap, or hardware utilization. Replacing the servers supporting a "sunsetting" application - an application on its way to retirement - wastes not only capital dollars, but also the operational expense needed to migrate it to the new hardware.
The most important aspect in the development and governance of a hardware refresh policy is finding the balance point between technical and cost considerations. Vital to the success of any IT hardware refresh policy is a specific application-by-application refresh rate that goes beyond the basic consideration of the age of the device. Optimizing IT hardware refresh cycles requires an understanding of the underlying technologies' past, present, and future use.
As a result, IT hardware refresh cycles often vary greatly across technology towers, driven by factors such as:
Using these drivers, a comprehensive policy will specify refresh cycles for the 5 major technology towers - server, storage, network, workstation, and security - as well as guidance on the categories within each technology tower (e.g., within servers, cycles for X86 and RISC architectures; within network, cycles for routers and switches).
A well-designed IT hardware refresh policy delivers a wide range of operational and financial benefits.
But capturing these benefits requires collaboration across the organization. The involvement of stakeholders from the engineering, operations, and application development groups in IT, and from Finance and key business users are needed if these benefits are to be fully realized.
Optimizing an IT hardware refresh policy starts with an analysis of the current IT hardware landscape. There are 5 steps to the analysis and the effort is usually led by the IT engineering group.
First, all IT hardware assets should be identified and reviewed. The fixed asset system and other internal asset databases are the usual starting points. However, in most large enterprises, this initial task can be extremely difficult, especially within the distributed computing environment or a company with a decentralized IT procurement process. When a central asset repository does not exist, analysis teams often work with operational and application teams (beginning with the most important applications) to do a "bottoms-up" discovery on the types and locations of hardware. This initial inventory is then compared to IT purchasing data provided by the procurement team or even from the Accounts Payable voucher report from the Finance team. When operational, procurement, and application teams cannot provide this information, analysis teams often go directly to their hardware vendors and ask them to provide detailed purchase information.
One immediate benefit of this inventory is that teams often discover hardware that is no longer being used - network switches in a storage room, decommissioned servers still powered up on the data center floor - but is still generating operating expenses in the form of hardware and software maintenance, and power and cooling costs. Clearing the books of the cost "zombies" is a valuable first step in for many IT organizations.
With the baseline inventory created, each piece of hardware is then categorized by tower and technology platform (e.g., server-X86, server-Sun, storage-SAN, storage-NAS). In the best case scenario, a configuration management database (CMDB) will have the hardware make and model information which can be easily mapped to each platform category. In less ideal situations, free-form description fields will require both custom queries and manual work to assign the inventory to the correct categories.
The next step is to identify each piece of hardware as active or inactive. Communication and management software can be used to identify if a hardware device is on the network - most teams assume that if a device is not on the network, it's probably not being used. Operational tools can measure utilization. But often, the most useful method involves the telephone or e-mail - scrubbing the inventory with the application and operational teams to identify the active servers that they own.
The inventory and procurement data for all the assets are then merged producing an age baseline of IT hardware by platform - determining the mean, median, and maximum age for each platform as well as the average life of retired assets. This age baseline provides important data that shows not only the age of the current equipment, but also the age of equipment that has been previously retired. This data is important in developing a fact-based understanding of a company's hardware usage practices.
Once the inventory of all in-scope assets is complete, an Incident Analysis is conducted to identify key performance benchmarks for each platform and category (e.g., average number of incidents per device by age). Keying off the IP address, host name, asset tag, or serial number in the baseline inventory and problem/change/incident management system(s), incident types and frequencies are correlated for each device by age and by platform type (see Figure 1). This analysis can be used to forecast when each type of platform will begin to see a drop-off in reliability. Combining this analysis with the cost of each incident develops a profile of overall operational costs by age of the platform - an important component of the CapEx-OpEx trade off analysis.
Another important input to the optimization of an IT hardware refresh policy is an understanding of the refresh policies and practices used by industry peers. Companies operating in a highly technology dependent industry where competitors are using cutting edge technology to provide superior performance and reliability should have a different refresh policy than companies in less technology-dependent industries. Using a third party familiar with technology practices within industries will provide the information necessary to complete this analysis.
An Operating Expense Impact Analysis is then conducted to identify operating expense costs for each platform and to support the development of the hardware refresh cost-benefit model. The analysis for both existing and potential replacement hardware should be on a total cost-of-ownership (TCO) basis, considering all impacts on operating expense - hardware depreciation, hardware maintenance, software maintenance, space, labor, HVAC, and power. In the example illustrated in Figure 2 below, the lower hardware maintenance and lower power costs of a new network switch create a situation where a new device has lower operating expenses than the older, fully depreciated network switch.
In addition to reducing operating expenses, significant hardware performance improvements can be realized through the proper and timely introduction of new devices. On a capacity and compute basis, hardware performance improvements are typically seen in the range of 30-50% and significantly contribute to the Operating Expense Impact Analysis.
Based on the findings of the Impact Analysis, a Technology Investment Roadmap can be developed to complement an updated IT Hardware Refresh Policy.
With the analysis completed and performed, and hardware refresh intervals established for each tower and technology category, a Hardware Refresh Decision Tree converts the hardware refresh policy into a path of Yes/No questions that allow IT staff to easily apply the policy in the day-to-day decisions they make across the enterprise.
In an example of using the decision tree, a baseline inventory aging analysis has been performed and shows that two of an application's eight servers have reached their recommended refresh age of three and a half years. The servers are still in use and the application is not scheduled for retirement. Working with the application team, you discover that these two servers are load balanced web servers. You also discover that these servers now have an average peak memory utilization of 80% and an average peak CPU utilization of 20%. Running an operating expense analysis, you find that by adding additional memory to your existing servers, you can meet your projected memory and CPU requirements and have lower operating expenses for the next two years than if you bought new servers. You make the decision to upgrade your existing servers by adding the additional memory. © 2009 Archstone Consulting. All rights reserved.
The execution of the refresh decision is not always straightforward. As enterprise technology landscapes become more complex and interdependent, the disposition of a single hardware component - the decision to maintain, upgrade, or replace -- must be considered at the application level. Modern architectures spread applications across multiple servers, storage and network devices, and data centers. Failure to consider refresh holistically through the analysis of application interdependencies, infrastructure interdependencies, and overall system level architectures can lead to painful and costly outages and customer dissatisfaction.
Just as executive support is instrumental in establishing an IT Hardware Refresh Policy, policy ownership and controls are necessary to maintaining one. An IT Hardware Refresh Policy should be owned by the IT Engineering group, but involves Finance, Application Development, and Operations stakeholders in the execution and upkeep of the policy.
Application owners know what is best for the business from a hardware and application perspective and should be given control over the implementation of the policy. To drive accountability, Finance Engineering, and Operations owners should monitor and enforce the policy. A RACI (Responsible, Accountable, Consulted, and Informed) model is often used to map roles and responsibilities across the groups. Some of the key roles and responsibilities include Engineering accountable for technology standards, Operations accountable and responsible for technology requirements, and Finance accountable for hardware refresh budgets.
Regular reviews of technology and application roadmaps and the creation of a hardware refresh calendar are the governance tools most often used to drive the application and upkeep of the refresh policy.
CIOs are being challenged to increase the business value of their company's existing investment in IT. Optimizing their IT hardware refresh policy - replacing accounting's useful life calendar with a more accurate set of useful technical lives - is a powerful but often-overlooked way to impact a significant portion of IT's capital and operating expense budgets.
With a rush of technology investment decisions on the horizon - broad implementation of server and storage virtualization, PC upgrades to handle the next-generation of Windows - CIOs that analyze their current infrastructure landscape and develop a hardware refresh roadmap provide their IT organizations with a framework to make decisions that drive the maximum business return out of tighter and tighter IT budgets.