Performancing Metrics

Vamsi Tokala's blog: Decoding concepts of Performance Engineering

Saturday, October 3, 2009

Decoding concepts of Performance Engineering

In most of the situations we have a tendency of using the formula’s without understanding the concepts behind it, because of which we may not be able to apply the laws correctly where it is required. 

The following example will help in understanding the some of the performance rules

A system was subjected to 10 requests in a span of 2 min and got the responses for all requests during this period and Utilization of the system during the time frame was observed to be at 50%.


Request Arrival rate for that system is  10/2=5 requests/Min
Throughput = 10/2 = 5 Responses/Min


in the above example, system utilized 50% of its resources for serving 5 requests. For Each request, it consumed 50%/5 of system resources. This value is called service demand of that resource


From the above example the formula for service demand can be derived as below
Service Demand  =Utilization/Throughput. 
similarly Utilization = Throughput * Service Demand This is also called utilization Law


The utilization of a resource is equal to the product of the throughput of that resource and the average service demand at that resource. 

2 comments:

Jackie Champion said...

Hey there! Keep it up! This is a good read. I will be looking forward to visit your page again and for your other posts as well. Thank you for sharing your thoughts about performance engineering. I am glad to stop by your site and know more about performance engineering.
One of the objectives of a peformance engineering is to eliminate late system deployment due to performance issues.
Software Performance Engineering is a critical component of Enterprise Risk Management. In today’s complex and interdependent system environments, technology must provide an outstanding user experience for ten’s of thousands users. The new workloads supported by these systems has become increasingly volatile with tremendous variance in average and peak. Peak workloads are no longer simply 3 or 4 times the average. They can rapidly increase to 10 or 20 times the average. Many systems in place today were not designed to rapidly scale to 20X the average. The performance and quality of software or a system is not just a technical issue, but also reflects how well the business operates and performs on a daily basis. In essence, proactive performance engineering is enterprise risk management. PE ensures systems are able to scale with unexpected spikes in system demand whether driven by marketing events, or today's more volatile financial market behavior.

steven said...

Yours guidence on Loadrunner is helpful to learners. Thanks for providing the needed information on Lordrunner online training