All quizzesHard
Reliability & Availability — Series 3
Preview — 3 of 10 questions
A request-based SLO permits 0.1% failures. The current failure rate is 2% over a representative interval. What is the error-budget burn rate?
A20 times the sustainable rate.
B2 times the sustainable rate.
C200 times the sustainable rate.
D0.2 times the sustainable rate.
Two replicas each historically fail 1% of the time, but both depend on the same power supply. Can their combined downtime probability be assumed to be 0.01%?
ANo; their combined downtime must always equal 2%.
BYes; provided the replicas use different process IDs.
CYes; replication makes failures statistically independent.
DNo; that multiplication assumes independent failures.
A load generator sends its next request only after the prior one completes. During a stall it sends far fewer requests. Why can this understate user-facing tail latency under a fixed arrival rate?
AIt guarantees every measured request bypasses the cache.
BIt omits arrivals that would have queued during the stall.
CIt changes every measured percentile into an average.
DIt doubles the server's actual available processing capacity.
Sign up free to play
Answer all 10 questions (7 more), see explanations for every answer, and track your score.