Queuing Theory: Little's Law
According to the Queueing Theory (Little's Law), if the average number of customers in the queue = L, the average arrival rate = A, and the average time a customer spends in the queue = W, then Little's Law says that L = A * W.
If we apply this same theory to Kanban software development, then we can say:
- WIP (work in progress) = Throughput * Cycle Time
- Or, Throughput = WIP/Cycle Time
- Or, Throughput ∞ 1/ Cycle Time (when WIP is constant)
- WIP = The average number of work items in the process
- Throughput = The average number of work items that are in the "done" state (i.e., average departure rate)
- Cycle Time = The average time a work item spends in the process (delivery date – start date)
That means that if we would like to increase the Throughput in our system, we'd have to reduce the Cycle Time of a work item.
Scenario 1: Calculation of WIP limit
Suppose that Throughput = 3 work items per day and Cycle Time per work item = 2 days, then WIP = 3 * 2 = 6. That means that only a maximum of 6 items are allowed to stay in the work-in-process field. This is called "limiting WIP."
Scenario 2: Comparison of Throughput based on constant WIP
Suppose the WIP is constant and equals to 2, and we observe that the Cycle Time we have per work item is 8 days, then Throughput = 2/8 or 0.25.
The team decides that it needs to increase its throughput from 0.25 to 0.5 to make a faster delivery. Then the "Cycle Time" calculation would be 2/0.50 = 4 days per work item.
Kanban provides greater visibility, perfect measurement, and faster feedback for improvements in the pull system by using this process.