Performance and Capacity
Little Law Calculator
Solve one of average work in progress, throughput, or cycle time from the other two.
Enter the values for Little Law
For Little Law, keep workload, resource boundary, units, and observation interval consistent.
Average Work In Progress and supporting Little Law values will appear here.
What Little Law calculates
Little Law answers one bounded performance or capacity question. Solve one of average work in progress, throughput, or cycle time from the other two. Its primary output is average work in progress, not a hardware ranking, service guarantee, or prediction about an unmeasured system.
Use Little Law for solving average work in progress from two consistently bounded observations.
A similar Little Law number from another benchmark version, host boundary, time window, or accounting convention may answer a different question.
Preparing a defensible Little Law case
The visible Little Law example begins with Average throughput = 18 items/s; Average cycle time = 2.4 s. Replace all defaults using measurements and assumptions from one coherent case.
Before Little Law, distinguish measured counters and rates from allocations, reserves, targets, and theoretical fractions. Label assumptions so they are not mistaken for observations.
Use matching time units and resource definitions in Little Law.
Arithmetic used by Little Law
The independent Little Law relationship is throughput × average cycle time. Supporting values expose the intermediate rate, ratio, count, headroom, or duration.
Carry unrounded values through Little Law.
Repeat Little Law in a spreadsheet or rearrange the equation when possible.
Reading the output from Little Law
Interpret Little Law with its numerator, denominator, and observation boundary.
When two Little Law cases differ, first compare workload, interval, success criteria, reserves, worker definitions, and whether values are measured or modeled.
The precision of Little Law cannot exceed its least certain input.
A controlled-input test for Little Law
Change one Little Law field and predict the output direction before recalculating. Restore it, then change a denominator, reserve, or worker count.
The simplest Little Law boundary is: Zero throughput or zero cycle time produces zero average work in progress. Test that case before trusting a large production-sized scenario.
If Little Law moves unexpectedly, inspect the first intermediate quantity and unit rather than adjusting an unrelated allowance.
Reverse-checking Little Law
Reverse the Little Law relationship where practical and see whether the original counter, rate, resource count, or duration returns.
For a whole-count Little Law result, test the immediately smaller count and confirm that it fails the stated capacity boundary.
Limits particular to Little Law
When auditing Little Law, little's Law requires consistent long-run boundaries for arrivals, departures, work in progress, and cycle time.
Little Law does not recommend hardware, predict benchmark scores, estimate unmeasured electrical power, diagnose a live system, or guarantee capacity and latency outcomes.
For Little Law, if contention, burstiness, skew, failures, warm-up, queue discipline, scheduler behavior, or workload variation matters but has no field, document it outside Little Law.
Recording Little Law reproducibly
A reproducible Little Law record includes raw counters, interval endpoints, workload identity, resource boundary, units, filters, software version, and measurement date.
Separate observed Little Law values from chosen targets, reserves, efficiencies, and theoretical fractions. The distinction determines what can be validated later.
Preserve prior Little Law cases rather than overwriting them.
Units and denominators in Little Law
Within Little Law, percentages retain their bases, rates retain their time units, and memory values retain their capacity or allocation definitions.
Do not mix decimal and binary memory quantities in Little Law without an explicit conversion.
During a Little Law check, for ratios above one, say which side is numerator.
Using Little Law in a capacity workflow
Pass Little Law to Cache Hit Ratio Target Calculator only with its unrounded value, units, timestamp, and boundary.
Compare the Little Law estimate with later observed behavior on the same workload. Retain the difference before changing reserves or model inputs.
Use Little Law as one auditable worksheet line alongside monitoring and workload evidence, not as a substitute for them.
Rechecking the visible Little Law example
Run Little Law with Average throughput = 18 items/s; Average cycle time = 2.4 s. Independently apply throughput × average cycle time and compare supporting quantities before the rounded output.
Replace one Little Law default at a time.
For Little Law, if a later observation differs, preserve both cases and inspect workload mix, interval, resource scope, averages, rounding, and excluded overhead.
Measurement quality in Little Law
The strongest Little Law input comes from a counter or timed observation collected across the exact workload boundary used in the denominator.
For a variable Little Law workload, retain more than the average.
Repeat the Little Law measurement under unchanged conditions before treating a difference as meaningful.
If the Little Law result supports planning, run a lower and upper observed case.
Reviewing the Average Work In Progress result in context
Start a review of Little Law with average throughput. For Little Law, confirm its unit, source, and observation date, then check that average cycle time describes the same population or system boundary. A familiar-looking magnitude is not enough when those definitions differ.
Create a second Little Law case rather than editing the saved result in place. Change one field, predict how average work in progress should respond from the written relationship, and compare the first supporting quantity that moves.
Questions about little law
Which inputs define Little Law?
Little Law uses Average throughput, Average cycle time. No live host, benchmark service, provider, or monitoring system is queried.
How can I verify Little Law?
For Little Law, repeat this relationship independently: throughput × average cycle time. Change one input and predict the direction before rerunning it.
What boundary matters in Little Law?
The Little Law inputs must describe the same workload, resource pool, interval, and accounting convention. Similar numbers from different boundaries should not be combined.
How should an unexpected Little Law result be checked?
Return to the saved inputs, vary average throughput alone, and compare the first supporting quantity that changes. This is more reliable than adjusting several fields until average work in progress looks familiar.