Performancing Metrics

Vamsi Tokala's blog: Little's Law

Monday, November 23, 2009

Little's Law

In continuation with my earlier post “Decoding concepts of Performance Engineering” I will try to derive Little's law in this post for better understanding

Consider a steady state system which has only one user and sends the requests at consistant intervals and response time of the requests were observed to be 1 sec.

No of Customers in the System = 1
Response Time = 1 sec (Service Demand)
So Maximum Throughput = (1/Service Demand) = 1 Request/Sec (Based on Utilization Law)

It implies average number of customers required being in the system to achieve 1 request/ sec with an average response time is 1 sec is 1 user

Similarly assume if the response time of the system is 0.1 Sec

Response time = 0.1
Max Throughput = 1/Service Demand = 10 requests/Sec

So One single user can simulate max of 10 requests/sec with a average response time of 0.1 sec

Little's Law

Average Number of Customers = Response time * Throughput

Little’s Law states that average number of customers in a system is equal to product of throughput and response time.

No comments: