System Architecture and Reliability

A cluster in an active-active configuration shares processing evenly among five servers. Utilization is proportional to the amount of processing. The goal is to keep each server's utilization at 80% or less even when one server fails and the remaining four take over its processing evenly. Which is the upper limit of each server's utilization in normal operation (all five running)?

  1. A48%
  2. B64%
  3. C80%
Show answer and explanation

Answer: B. 64%

Let u be each server's utilization in normal operation; taking one server's capacity as 1, the total amount of processing is 5u. When one server fails, the remaining four share it, so each server's utilization becomes 5u ÷ 4, and keeping this at 80% or less requires u ≤ 80% × 4 ÷ 5 = 64%. In an active-active configuration every server does processing, so enough spare capacity to take on a failed server's share must be kept even in normal operation.

  • A. This is the value when two servers fail and three take over (80% × 3 ÷ 5).
  • B. Correct. 80% × 4 ÷ 5 = 64%.
  • C. This value does not consider taking over the failed server's share.
Practice this topic Back to topic

Related calculators

Books on System Architecture and Reliability Ad

As an Amazon Associate I earn from qualifying purchases.

More System Architecture and Reliability questions

  1. A system uses three servers with an availability of 0.95 each and has the required capacity as long as at least two of them are running. The servers fail independently, and failures of devices other than the servers are ignored. Which is the availability of this system?
  2. Which is an example of a design based on the fail-soft concept?
  3. In semiconductor manufacturing equipment, normally closed gas supply valves that open with air pressure are used, so that if a power outage or a loss of compressed air occurs, the valves close and the supply of hazardous gas stops. Which design concept is this?
  4. Which is an example of a design based on the fool-proof concept?
  5. Among the approaches to improving system reliability, which corresponds to fault avoidance?