Monitoring cloud resources
The metrics the platform provides, the period and the statistic, the alarm and its three states, health checks, saturation forecasting and cost budgets.
- 24 cards
- 6 topics
- Free
The outline, topic by topic
Each topic is weighted the way the exam weighs it.
- Concepts
- Questions
- Sort and link
- The numbers
- The curves
- Scenarios
A few cards from the deck
Metric: what is it?
Show the answer
A series of numeric measurements of a resource, taken at regular intervals: CPU, memory, disk operations, network traffic.
Each point is aggregated over a period with a statistic (average, maximum, percentile): the same series tells different stories depending on that choice.
In which state is an alarm whose metric no longer arrives?
- Insufficient data
- OK, since nothing exceeds the threshold
- ALARM, since a measurement is missing
- Disabled, until the instance comes back
Show the answer
Insufficient data
Without measurements the condition can be evaluated neither true nor false. It is often the sign of a dead instance.
An average hides extreme cases; to watch the worst users experience, look instead at the 95th or 99th …, or at the ….
Show the answer
An average hides extreme cases; to watch the worst users experience, look instead at the 95th or 99th percentile, or at the maximum.
If 1 request in 100 takes ten seconds, the average stays good while the 99th percentile reveals it.