Performance Concepts: Difference between revisions
Jump to navigation
Jump to search
| Line 6: | Line 6: | ||
=Throughput= | =Throughput= | ||
Throughput is the rate at which something can be produced, consumed or processed, in a time unit. Throughput is usually relevant in case of [[System_Design#Batch_Processing|batch processing systems]], such as Hadoop. | |||
=Latency= | =Latency= | ||
Revision as of 18:32, 8 November 2021
Internal
Throughput
Throughput is the rate at which something can be produced, consumed or processed, in a time unit. Throughput is usually relevant in case of batch processing systems, such as Hadoop.
Latency
- review How NOT to Measure Latency https://www.infoq.com/presentations/latency-response-time
Latency and response time are synonymous: the length of time it takes something interesting to happen.
Standard deviation does not have any meaning for a dataset that describes latency. It is not relevant.
Latency must be measured in the context of load, measuring the latency without load is misleading.
Define saturation.
Identify where the saturation point is. Don't run a system at saturation or over.
Queueing Theory
https://en.wikipedia.org/wiki/Queueing_theory
- Response Time in Queueing Theory.
- Service Time in Queueing Theory.
Response time and service time diverge as saturation becomes worse.
Organizatorium
- Don't censor bad data, don't throw away data selectively.
- Never average percentiles.
- Coordinated omission. Coordinate omission usually makes something that you think is a response time metric only represent a service time component.
- Response Time in Queueing Theory.
- Service Time in Queueing Theory.
- xth percentile (quantiles) - the value of the performance parameter at which x% of the request are better; https://www.vividcortex.com/blog/why-percentiles-dont-work-the-way-you-think