1.what is standard deviation? 2.what is...

  • Thread starter Ankammarao Maddineedi
  • Start date

  • perf-test.com need your contributions to build up a strong repository of performance engineering resources.

A

Ankammarao Maddineedi

Guest
1.what is standard deviation? 2.what is lattency? 3.what is 90th percentile time, what is the use of that and how to caluclate 90th percentile time..? 4.what is IP spoofing ? 5.Whar is littleslaw..? Pz give exact and esay answers for these questions
 
yes tried but i am not getting clear idea
 
Please research urself and post answers here. we will help you to get more clarity.
 
Hi, ip spoofing means it will allocate different IPS for every vuser, need to enable ip spoofing in controller 90th percentile =total passed transactions*90/100 Plz share remaining questions answers
 
ok ok ok, I will answer a bit ;) Standard Deviation is mathematical term/number for how much the set of measurements deviate from the average. The lower the standard deviation, the more accurate the average is. The higher the deviation, the lesser the average tells you Latency (network latency for this matter) means the time it takes for one network packet to arrive on the endpoint. For example, a single packet takes up to 40/50 ms to travel from Europe to the US. The higher the latency, the less responsive something can be. Imagine a website with tons of static objects, it might take a long time to communicate and receive all object. The 90th percentile means the 90% number/measurement. If you order all numbers/measurements, lets make it easy: 100 numbers, from low to high, the 90th percentile means the 90th number. With performance testing you will have peaks most of the time and this will ruin your average. Therefor the 90th percentile can be used. ONLY use 90th percentile if you have tons of measurements. IP-Spoofing means you use different IP-addresses than the main IP-address of the loadgenerator. If a loadgenerator has lets say 100 IP-addresses assigned, you can distribute the IP-addresses over your virtual users simulated from that loadgenerator. A very important feature if you test over the internet, because else you will be limited to a number of IP-addresses of the number of loadgenerators. Little's Law... well, it's just guessing an average response time according to throughput. Just don't bother this, use common sense for what you measure. With parallel processing you cannot guess response times, as throughput can be high, but response times can be high too.
 
nice explanation Bro........ Joerek van Gaalen.....but,thinking that there is no any directly proportion relation between response times and throughput until the system resources are fine.when we increase the users,hits will increase,when hits increases obviously throughput will increases,but until system resources are fine response times will be under SLA....if resources are exhausted then only response times going to spike/ cross SLA/increases..give me like if you agree else give me the explanation bro...just wanted to know about this.
 
You must log in or register to reply here.
Top Bottom
Back