Struct WorkloadGenerator 

pub struct WorkloadGenerator { /* private fields */ }

A Zipfian workload generator for cache benchmarking.

This structure generates a fixed set of random keys and samples them according to a Power Law distribution. This mimics real-world scenarios where a small subset of “hot” keys receives the majority of traffic.

Implementations

impl WorkloadGenerator

pub fn new(count: usize, skew: f64) -> Self

Creates a new workload with a specified number of unique keys and skew.

Arguments

• count - The number of unique logical keys in the workload universe.
• skew - The Zipfian exponent ($s$).
  • 0.5: Approaching uniform (flatter distribution).
  • 1.0: Standard Zipfian (classic “Power Law”).
  • 1.5: Highly skewed (very few keys receive almost all traffic).

Panics

Panics if skew is not within the statistically significant range of 0.5 to 1.5.

pub fn key<D: Rng>(&self, distribution: &mut D) -> Bytes

Samples a key from the workload using the provided random number generator.

This method is designed for high-concurrency environments; by passing a mutable reference to a thread-local Rng, it avoids global lock contention.

Performance

This operation is $O(1)$ relative to the size of the key strings but follows the complexity of the rand_distr::Zipf sampling algorithm (typically rejection inversion).