Skip to content
Tutorials Center
My learning

Topics

Lesson 11 of 38 · M/M/1 Queues

M/M/1 Performance Measures

Derive L, Lq, W and Wq for the M/M/1 queue from its steady state, verify them with Little's law and put them in a reusable calculator.

With the steady-state distribution Pn = (1 - ρ) ρ^n in hand, the performance measures follow by simple sums and Little's law. These four formulas are worth memorizing, because they are the baseline against which every other queueing model is compared.

Deriving L

The mean number in the system is

L = Σ n Pn = (1 - ρ) Σ n ρ^n

For |ρ| < 1 the sum Σ n ρ^n equals ρ / (1 - ρ)², so

L = ρ / (1 - ρ)

You can also see this as the mean of a geometric distribution that starts at zero: (1-ρ) is the success probability, so the mean number of "failures" is ρ/(1-ρ).

The other three by Little's law

  • Queue length. The number in the queue is N - 1 when N ≥ 1, and 0 otherwise. So Lq = L - ρ = ρ² / (1 - ρ).
  • Time in system. Little's law L = λW gives W = L/λ = 1 / (μ - λ).
  • Time in queue. Wq = W - 1/μ = ρ / (μ - λ), which equals Lq/λ.
Measure Formula
L ρ / (1 - ρ)
Lq ρ² / (1 - ρ)
W 1 / (μ - λ)
Wq ρ / (μ - λ)

Take a help desk that gets 8 requests per hour and resolves 10 per hour. Then ρ = 0.8, L = 4, Lq = 3.2, W = 1/(10 - 8) = 0.5 hours (30 minutes) and Wq = 0.8/2 = 0.4 hours (24 minutes). Notice that 24 of the 30 minutes are spent waiting. Only 6 minutes are service.

A symbolic check

A computer algebra system can confirm the sum for us. Here we use SymPy to compute Σ n (1-ρ) ρ^n for the concrete value ρ = 4/5, then get Lq as L - ρ.

Python Check L with SymPy

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

A reusable calculator

The function below returns all measures for an M/M/1 queue and checks that they satisfy Little's law and the identity L = Lq + ρ. Use the same time unit for λ and μ.

Python M/M/1 calculator

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

How the measures scale

Compare the formulas at different utilizations. With μ = 1 per minute, each measure is a function of ρ alone. They all blow up like 1/(1-ρ). For example L at ρ = 0.5, 0.9, 0.99 is 1, 9, 99.

Python The 1/(1-rho) blow-up

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

A useful rule: halving the slack 1 - ρ doubles the queue. Going from 90% to 95% busy halves the slack from 0.10 to 0.05, so the waits double.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. An M/M/1 queue has lambda = 3 and mu = 4 per minute. What is L?

    rho = 0.75 and L = rho / (1 - rho) = 0.75 / 0.25 = 3.
  2. 2. With lambda = 3 and mu = 4 per minute, what is W?

    W = 1 / (mu - lambda) = 1 / 1 = 1 minute.
  3. 3. What is Lq for the same queue?

    Lq = L - rho = 3 - 0.75 = 2.25, or rho^2/(1-rho) = 0.5625/0.25 = 2.25.
  4. 4. If rho goes from 0.9 to 0.95, L changes from 9 to ...

    L = 0.95 / 0.05 = 19, which is more than double.

Summary

  • L = ρ/(1-ρ), Lq = ρ²/(1-ρ), W = 1/(μ-λ), Wq = ρ/(μ-λ).
  • They are linked by Little's law and by L = Lq + ρ, W = Wq + 1/μ.
  • All blow up like 1/(1-ρ) as utilization approaches 1.
  • Always check units and that λ < μ.

Next: The Waiting-Time Distribution