How this instrument works
The M/M/1 model describes a single-server queue where customers arrive randomly at a steady average rate λ (arrivals per hour) and are served, one at a time, at a steady average rate μ. Three figures describe its long-run, steady-state behavior: utilization, ρ = λ⁄μ, the fraction of time the server is busy; the average number of customers in the system (waiting plus being served), L = ρ⁄(1−ρ); and the average time a customer spends in the system, W = L⁄λ.
The formulas share a common warning built into their structure: as the arrival rate approaches the service rate (ρ approaches 1), the denominator (1−ρ) shrinks toward zero, and both the average queue length and the average wait grow without bound. A server running at 99% utilization doesn't perform merely a little worse than one at 50% — it can develop an arbitrarily long queue, the reason real systems are almost always designed with deliberate slack rather than running at their theoretical maximum capacity.
This same M/M/1 structure describes any single-server system where arrivals and service times are both reasonably random and memoryless: a help-desk ticket queue, a single checkout lane, a print job spooler, or a single-threaded task queue in software — the underlying arithmetic doesn't care what's actually being served.
- Enter the average arrival rate (customers, tasks, or requests per hour) into the Arrival rate (λ) field.
- Enter the average service rate into the Service rate (μ) field.
- Read Utilization, Average number in system, and Average time in system: the sheet computes all three from the M/M/1 formulas.
Worked example — arrivals at 4, service at 5 per hour
Customers arrive at an average of 4 per hour, and the single server processes them at an average of 5 per hour. Utilization is 4⁄5=80%, the average number in the system is 0.8⁄0.2=4, and the average time in the system is 4⁄4=1 hour — a busy but stable queue, with the server working most of the time but the wait staying finite.
A lightly loaded version of the same system, arrivals at 2 and service at 10, drops utilization to just 20%, the average number in the system to 0.25, and the average wait to just 7.5 minutes — comfortable slack, a queue that rarely builds up at all.
Questions
What does M/M/1 mean?
It's shorthand for a queue with Markovian (memoryless, exponentially distributed) arrivals, Markovian service times, and 1 server — the simplest and most widely used queueing model, describing a single-server line with random arrival and service timing.
What happens as utilization approaches 100%?
Both the average queue length and the average wait grow without bound, since the formulas divide by (1−ρ), which shrinks toward zero. A server running near its theoretical capacity can develop an arbitrarily long queue, not just a modestly longer one.
Why must the arrival rate be less than the service rate?
If arrivals came in faster than the server could process them, the queue would grow forever with no steady state to describe — the M/M/1 formulas only apply to a stable system where the server can, on average, keep up.
Does this model apply outside of literal customer queues?
Yes — any single-server system with reasonably random arrival and service timing fits the same structure: a help-desk ticket queue, a print job spooler, a single-threaded software task queue, or a single checkout lane all obey the identical M/M/1 arithmetic.
What is the difference between L and W?
L is the average NUMBER of customers in the system at any given moment (a count); W is the average TIME any one customer spends in the system (a duration). They're related by Little's Law, L = λW, which this calculator applies directly.