Queuing Theory Calculator | M/M/1 & M/M/c Queue Metrics
Model M/M/1 (single-server) and M/M/c (multi-server) Poisson arrival queues. Computes server utilization ρ, average number in system L, average queue length Lq, mean system time W, mean wait time Wq, and probability P(n) of exactly n customers.
What Is the Queuing Theory Calculator | M/M/1 & M/M/c Queue Metrics?
Queuing theory models systems where customers arrive, wait for service, and depart. The M/M/1 model assumes Poisson arrivals at rate λ, exponential service at rate μ, and a single server. The M/M/c model extends this to c parallel servers, each with rate μ. Stability requires ρ = λ/(c·μ) < 1 — if arrivals exceed total service capacity, the queue grows unboundedly. Little's Law, L = λ·W, relates mean queue length to mean system time and holds for any stable queue regardless of distributional assumptions.
Formula
M/M/1: ρ = λ/μ, L = ρ/(1−ρ), Lq = ρ²/(1−ρ), W = 1/(μ−λ), Wq = ρ/(μ(1−ρ)) · M/M/c: ρ = λ/(cμ), Erlang C = P₀·(cρ)ᶜ/[c!(1−ρ)], Lq = C(c,λ/μ)·ρ/(1−ρ) · Little's Law: L = λ·W
How to Use
- 1
Enter the arrival rate λ — the average number of customers arriving per hour (e.g., 8 customers/hr at a bank).
- 2
Enter the service rate μ per server — the average number of customers one server can handle per hour (e.g., 10/hr).
- 3
Enter the number of servers c (1 for single-server M/M/1, or 2–20 for multi-server M/M/c).
- 4
Check the stability condition: the system is stable only if ρ = λ/(c·μ) < 1.
- 5
Click "Analyze Queue" — see server utilization ρ, mean customers in system L and queue Lq, mean times W and Wq.
- 6
Review the Erlang C probability (M/M/c): the probability that an arriving customer must wait.
- 7
Check the P(n) table for the probability of exactly n, at least n, or at most n customers in the system.
Enter the arrival rate λ (customers per hour), service rate per server μ (customers per hour), and number of servers c. Set c = 1 for M/M/1, or c ≥ 2 for M/M/c. Click Analyze Queue to see all performance metrics including utilization, mean queue length, mean waiting time, and the probability distribution P(n).
Example Calculation
Bank teller (M/M/1): λ = 8 customers/hr, μ = 10/hr, c = 1. ρ = 0.8. L = 0.8/(1−0.8) = 4 customers. Lq = 0.64/0.2 = 3.2. W = 1/(10−8) = 0.5 hr = 30 min. Wq = 0.8/(10×0.2) = 0.4 hr = 24 min. Call center (M/M/5): λ = 50/hr, μ = 12/hr, c = 5. ρ = 50/(5×12) = 0.833. Erlang C ≈ 0.68. Lq ≈ 3.4. W ≈ 5.4 min. Adding a 6th agent reduces Erlang C dramatically.
Understanding Queuing Theory | M/M/1 & M/M/c Queue Metrics
M/M/1 vs M/M/c Key Formulas
| Metric | M/M/1 | M/M/c |
|---|---|---|
| Utilization | ρ = λ/μ | ρ = λ/(cμ) |
| Stability condition | ρ < 1 | ρ < 1 (per server) |
| P(0) — system empty | P₀ = 1 − ρ | Erlang formula (see below) |
| P(wait) — customer waits | P(wait) = ρ | Erlang C = C(c,λ/μ) |
| Avg queue length Lq | ρ²/(1−ρ) | C(c,λ/μ)·ρ/(1−ρ) |
| Avg system length L | ρ/(1−ρ) | Lq + λ/μ |
| Mean wait time Wq | ρ/(μ(1−ρ)) | Lq/λ |
| Mean system time W | 1/(μ−λ) | Wq + 1/μ |
| P(n in system) | (1−ρ)ρⁿ | More complex (see formula) |
Effect of Server Utilization on Queue Length (M/M/1)
| Utilization ρ | Avg in system L | Avg in queue Lq | Mean system time W (×1/μ) | Interpretation |
|---|---|---|---|---|
| 0.50 | 1.00 | 0.50 | 2.0 × | Low load — quick service |
| 0.70 | 2.33 | 1.63 | 3.3 × | Moderate — manageable waits |
| 0.80 | 4.00 | 3.20 | 5.0 × | High — noticeable queue builds |
| 0.90 | 9.00 | 8.10 | 10 × | Very high — long unpredictable waits |
| 0.95 | 19.0 | 18.1 | 20 × | Near saturation — unacceptable queues |
| 0.99 | 99.0 | 98.0 | 100 × | System near collapse |
Real-World Applications
- ›Call center staffing: Erlang C determines how many agents to staff so that, say, 80% of calls are answered within 20 seconds. Standard workforce management software uses M/M/c as its core model.
- ›Hospital emergency departments: Patient arrivals are approximately Poisson; M/M/c models help size triage staff. Multi-priority models (M/M/c/K) account for urgent vs non-urgent patients.
- ›Network packet queuing: Internet routers model traffic as Poisson arrivals. Packet delay and loss probability at a router follow M/M/1 or M/D/1 (deterministic service) queuing models.
- ›Supermarket checkout: Retailers use M/M/c to decide how many checkout lanes to open based on time-of-day arrival rates, balancing cashier cost against customer wait time.
- ›Manufacturing WIP: Factory workstations model as queues. By Little's Law, reducing WIP (Lq) reduces cycle time (Wq). This is the basis of lean manufacturing and the Theory of Constraints.
Frequently Asked Questions
What does server utilization ρ mean?
Utilization ρ = λ/(c·μ) is the fraction of time each server is busy. ρ = 0.8 means servers are busy 80% of the time. When ρ approaches 1, queue length and waiting times grow toward infinity — even small fluctuations cause large backlogs. In practice, real systems should target ρ < 0.85 to maintain manageable queue lengths.
What is Little's Law and when does it apply?
Little's Law states L = λ·W: the average number of customers in a stable system (L) equals the arrival rate (λ) times the mean time each customer spends in the system (W). It applies to any stable queuing system regardless of arrival distribution, service distribution, or number of servers — making it remarkably general. It also applies separately to the queue alone: Lq = λ·Wq.
What is Erlang C and how is it used?
Erlang C (the M/M/c waiting probability) is the probability that an arriving customer must wait because all c servers are busy. It is the key output of call center staffing models. If Erlang C = 0.30, 30% of callers wait. Service level targets (e.g., 80% of calls answered within 20 seconds) are calculated from Erlang C. Higher c lowers Erlang C dramatically near saturation.
What are the M/M/1 queue assumptions?
The M/M/1 model assumes: Markovian (Poisson) arrivals, Markovian (exponential) service times, 1 server, infinite queue capacity, infinite customer population, and first-come first-served discipline. The first M refers to the Markovian arrival process, the second M to Markovian service, and the 1 to the number of servers.
How do I decide how many servers to add?
Compare the cost of adding a server against the cost of customer waiting. Use the M/M/c calculator: try c=1, 2, 3, ... and observe how Lq, Wq, and Erlang C change. There is typically a steep improvement in queue metrics when adding servers near saturation (ρ close to 1), with diminishing returns as ρ drops below 0.7. The optimal c balances server idle cost against customer wait cost.
You Might Also Like
Explore 360+ Free Calculators
From math and science to finance and everyday life — all free, no account needed.