Donelan.COM
Data Center Design Benchmarks - General
The design and operation of Internet data centers is the art of compromise. One way to organize the choices is categorizing the requirements in to different levels.
| Level I | Level II | Level III | Level IV | Level V | |
| Examples | Single site data processing facility, limited off-site impact due to failure | End-office telecommunications facility, localized impact due to failure | Toll-office telecommunication facility, regional impact due to failure | High-end financial or airline communication facility, national impact due to failure | Critical communications center, One-of-a-kind buildings (e.g. White House, Pentagon, etc), international impact due to failure |
| Typical Size (substitute measurement for "importance") | Less than 2,500 square feet | 2,500 to 80,000 square feet | 80,000 to 150,000 square feet | more than 150,000 square feet | Specialized requirements |
| General design | Meets local code. System is able to withstand the temporary loss of off-site commercial utility power. | System is able to withstand a failure of a single active component with minimal impact to critical loads. | Meets code requirements for essential facilities. System is able to withstand a failure of a single active component. Planned maintenance without impacting any critical load. Fault tolerance may be reduced during maintenance activity. | Systems are able
withstand a single, worst-case infrastructure failure without impacting any
critical load. Planned maintenance activity without impacting any critical load. |
Design for maximum probable event. Systems are capable of supplying services continually during any scheduled or unscheduled, natural or man-made disruption. |
| Program elements | The design shall eliminate or reduce the probability of events that can cause injury or death to personnel or damage to or loss of equipment or property. Systems of high complexity or peculiarity shall be identified and minimized. | Standardized design to
minimize equipment and component stockage. Redundant systems shall be of the
same design. Individual system and subsystem test and checkout requirements shall be developed to ensure safe and normal operation of the system. |
Reliability,
availability and maintainability requirements shall be implemented during
design to maximize the availability of the systems. |
Consolidated systems test program covering all phases of testing to develop confidence in the systems and provide for interim and final acceptance of equipment and complete systems. | Human factors
engineering will ensure that reliability, availability and safety are not
degraded through human activities during operation or maintenance. The design incorporates, within program constraints, the highest level of inherent safety. |
| Reliability, availability and survivability | 99.8000% | 99.9000% | 99.9900% | 99.9990% | 99.9999% |
| Real-world availability (5 years) | 99.7500% | 99.8500% | 99.9000% | 99.9910% | 99.9990% |
| External design threat considerations | Unscheduled loss of commercial power | Component failure, severe weather | Local disaster (tornado, small earthquake), intrusion, accidental human error | Regional disaster (hurricane, earthquake), bad publicity, individual attack | Insider threats, national disaster, war, civil unrest |
| Life-cycle cost analysis | Initial cost is the primary factor | 5 to 10 year occupancy, 1 year or less payback period | 10-20 year occupancy, 1-3 year payback period controlled by depreciation | 10-20 year occupancy, 5-10 year life cycle cost analysis | Permanent occupancy, 25
years or longer. Total life cycle cost analysis |