Although hardware price/performance continues its ride down the commodity curve, IT organization (ITO) maintenance budgets remain stubbornly high. Users focusing on lean maintenance management should realize significant savings against an often-ignored budget line item.
META Trend: Sticking to a regimen of continuous process and infrastructure rationalization through 2008, operations (build and run sections) will drive service and product optimization, providing 5%-10% annual unit-cost improvements via improved hardware utilization, along with maturing operational process and asset management. Moreover, maturing virtualization plus workload management and partitioning - coupled with requisite provisioning and infrastructure/application management software and driven in part by rational consolidation - will enable platform utilization to more than double during the next five years, providing needed flexibility for expanded adaptive service delivery. For the 70% of the Global 2000 that are successful at transforming to "flexible manufacturing," strong process and improved infrastructure ROI will propel operations toward the role of an agile, trusted partner.
For more than 10 years, the reliability of high-end systems - both servers and disk storage - has increased dramatically. Driven by user demand, aggressive competition, and the reliability inherent in component miniaturization, vendors have been able to improve component reliability by 40%-50% per year. Moreover, increasingly sophisticated soft-fail diagnostics, redundancy, and transparent autonomic self-healing technologies have masked any remaining component reliability issues to deliver very high availability of the overall server or storage subsystem (see Figure 1).
However, the price of maintenance has not kept pace with such dramatic improvements in component reliability and subsystem availability. Fundamentally, the price a vendor charges to maintain a hardware asset has little relation to the vendor’s costs. Most infrastructure manufacturers view hardware maintenance more as a high-margin annuity than as a cost of doing business. In addition, inflated maintenance price structures are often instrumental in encouraging ITOs to accelerate refresh cycles on IT assets that could otherwise remain viable for 12-24 additional months (this is especially true in disk storage). There are several quick hits that the ITO should examine for rapid hard-dollar returns.
On the server side, most Intel-based servers have adopted the three-year warranty that has been standard practice for more than five years for disk storage subsystems. However, Unix servers have been slower to offer such an extended warranty, though competitive cracks are appearing. ITOs are encouraged to request a three-year warranty as a final net benefit to signing a competitive bid (if introduced too early in negotiations, the vendor will typically wrap the extended warranty’s cost into the asset’s purchase price). ITOs that standardize on three-year warranties are cautioned against complacency. They should also get specific caps on post-warranty maintenance to make a fourth or even fifth year economically feasible. Such caps must be aggressive to compete with steep price/performance (P/P) curves on the base hardware, which are declining 30%-35% per year. For example, a 35% annual P/P curve will enable the ITO to buy a new server for less than 20% of the current market rate within four years - which means post-warranty maintenance costs on current servers must be limited to approximately 5%-6% of current purchase price.
During the next three to five years, we project that storage and server hardware maintenance will largely disappear from the user budget as warranties are extended to match financing terms. This is already the case for most storage subsystems (three-year refresh cycles matching both finance and warranty terms) and is becoming standard practice for Intel-based servers as well. Thus, ITOs should require Unix-based systems to match this growing industry standard of a three-year warranty as a no-charge benefit to competitive procurements.
What Is the Current Hardware Maintenance Spend?
Such a basic question is often difficult for even the most sophisticated ITOs to answer quickly or accurately. Indeed, when challenged to quantify just their storage maintenance charges, many large users found they consume as much as 20%-35% of annual storage spend (it is not unusual for a storage vendor’s maintenance charges for three-year-old equipment to be double what it would cost to replace the equipment). One of the first places to look for maintenance-budget savings is the existing infrastructure portfolio. ITOs paying to maintain disk storage assets more than three years old can almost certainly realize substantial savings by replacing/consolidating them with new high-capacity subsystems with a three- to four-year warranty. Moreover, a fourth year of warranty will enable upgrades to remain affordable through the first 24+ months of the asset’s life. For example, with post-warranty storage maintenance costs often prohibitively high, a three-year warranty typically dictates a three-year life. As upgrade costs must be amortized over the asset’s remaining term, this three-year refresh cycle makes upgrades after 12-18 months financially unaffordable.
Along with a basic maintenance inventory, ITOs should categorize their infrastructure portfolios according to required maintenance levels. Specifically, ITOs should deploy a granular maintenance strategy that is mapped to the asset’s business criticality (e.g., 24x365; 8:00 am to 5:00 pm, Monday through Friday; time and materials). Although such monitoring can become counterproductive if taken to the extreme, identifying several gross categories of maintenance service and consequent costs can yield significant savings over the often automatically subscribed full 24x365 plan.
Managing Maintenance Risk: Third-Party Maintenance
In the consumer appliance and electronics industry, retailers can typically make more profit on selling extended warranties than on the initial sale of the base asset. Moreover, there exist generic insurers of such maintenance risk that will take on the actuarial risk for a set percentage fee. The insurer uses the asset’s actual maintenance track record to determine actuarial risk and consequent extended-warranty pricing. For IT assets, third-party maintainers (TPMs) can act in a similar capacity, lowering the ITO’s costs by taking on maintenance risk as a contractor. The TPM maintains most of a given IT portfolio and subcontracts back to the manufacturer the less profitable assets or those for which the TPM lacks skills or access to low-cost parts.
Our research indicates that robust, credible hardware maintenance competition will typically reduce the incumbent vendor’s original bid by more than 20% or save more than 40% by outsourcing to a reliable TPM. For maintenance of large diversified infrastructure portfolios, ITOs should limit competitive bids to TPMs with a credible global presence (e.g., IBM Global Services, Fujitsu, StorageTek, Hitachi). Asset distribution is another key variable in the TPM risk/reward equation. IT assets that are more centralized (e.g., within a large data center) are better candidates for competitive bidding than those thinly populated across geographies. Users should begin well in advance (12-24 months) of current maintenance-contract expiration to explore alternative suppliers.
Bottom Line: Hardware maintenance is an often overlooked (and unknown) but significant IT budget expense. ITOs that deploy rigorous portfolio and asset management techniques to hardware maintenance costs can harvest significant (20%-40%) cost savings and develop a more agile infrastructure.
Business Impact: Tighter controls over hardware maintenance expenditures will yield more efficient asset management and a higher return on the IT asset portfolio.
META Group originally published this article on 11 March 2004.