Wilde’s ability to meet the tight deadline requirements prevented programme delays (and therefore additional cost) which contributed to the success of the project. The report also contained suggestions to further improve the system in respect to the availability, reliability, and maintainability ensuring a more robust solution for the client.
Data Centre Design: Reliability, Availability, and Maintainability (RAM) Analysis | Reliability Blog
Business Benefits
Data Centre Background
For the past few years, Wilde have supported Clients who specialise in the design, development, and construction of prestigious Data Centres. Each Data Centre is different in configuration, complexity and location and the availability (or uptime) targets varied from Tier 1 (99.671% availability) to Tier 5 (99.999% availability). Tier 5 equates to a downtime of 5 minutes 15 seconds in a year.
The 3 major principles to ensure high availability are:
- No single points of failure
- Reliable crossover/failover points
- Failure detection capabilities
Project Aims:
- To undertake an ARM analysis of the electrical and mechanical services of the building.
- To qualify the design based on a “five-nines” inherent availability target.
- To suggest improvements for further developing the design in relation to RAM.
- To improve internal client reliability knowledge.
Requirement
Having previously collaborated on various projects, The Client approached Wilde Analysis to assist them with quantifying the availability of the new data centre. The desired objective was to achieve an aspirational availability target of 99.999%, for the combined mechanical services and electrical power distribution.
Due to the continuous demand on the data centre, it is essential that it is highly available. This design had additional complexity due to a changing system configuration at the start of the operational phase. As the power demands were scheduled to increase over a specific period, plans were formulated to introduce supplementary backup power supply to ensure optimal redundancy. The assessment encompassed all four distinct configurations.
Solution
Wilde Analysis engineers worked with the client to create a formal system definition to contain details of the system components, boundaries, and interfaces. They then created a system reliability block diagram (RBD) model using the modelling package ReliaSoft BlockSim. Within the model it was then possible to identify components from the system which contribute to failure, assign failure distributions and repair rates to these components, define the reliability wise configuration of the system and finally calculate the system availability using Monte Carlo simulation.
Wilde’s experience of reliability analysis and using the ReliaSoft packages enabled the provision of guidance and reassurance that the system was mapped correctly and the information output from the simulations fundamentally made sense.
Results
Wilde Analysis delivered to the client, on time, a report containing the ARM analysis and the evidence that the 99.999% availability targets could be successfully met for all four system configurations in this design. This allowed the client’s design to be accepted by their customer and for the project to progress.