TailEventRateScenario
Serving scenarios · Implementation: mlsysim/engine/scenario_definitions/serving.py.
Construct this scenario with keyword inputs. The fields below show the types and defaults declared by this class.
How often an event of probability event_probability per request occurs at request_rate.
Notes
Equation: Expected events per second=request_rate*event_probability.
Assumptions: A constant mean request rate and a probability applicable to each request.
Excludes: Event clustering, confidence intervals, and severity of an event.
References
Derivation: Linearity of expectation gives the mean number of qualifying requests as request count times the per-request event probability. Applying the same product to the request rate gives events per second without assuming independent event timing.
Examples
The named registry scenario supplies the stated inputs; this is a modeled calculation, not a new measurement.
>>> from mlsysim import Scenarios
>>> scenario = Scenarios.RecommenderTailEvents
>>> round(scenario.events_per_second, 6)
1000.0Attributes
| Name | Description |
|---|---|
| event_probability | |
| events_per_second | Expected event frequency in 1/s: request arrival rate times event probability; this is a mean, not a tail-latency bound. |
| request_rate |
request_rate
Type: Quantity.
Declared default: required.
event_probability
Type: float.
Declared default: required.
events_per_second
engine.scenario_types.TailEventRateScenario.events_per_second
Expected event frequency in 1/s: request arrival rate times event probability; this is a mean, not a tail-latency bound.
Equation: (self.request_rate * self.event_probability).to(‘1/second’).magnitude