What is 90th Percentile Response Time?

The 90th percentile response time value is the value for which 90% of the data points are smaller and 10% are bigger.

To calculate the 90^th percentile value:

1.       Sort the transaction instances by their value.

2.       Remove the top 10% instances.

3.       The highest value left is the 90^th percentile.

Consider the below example:

There are 10 instances of transaction “TRANSACTION 01” with the values

4 sec

2 sec

3 sec

1 sec

5 sec

17 sec

8 sec

7 sec

6 sec

10 sec

1.       Sort values from best to worst:

1 sec

2 sec

3 sec

4 sec

5 sec

6 sec

7 sec

8 sec

10 sec

17 sec

2.       Remove top 10%, in our case i.e. value “17 sec”

1 sec

2 sec

3 sec

4 sec

5 sec

6 sec

7 sec

8 sec

10 sec

3.       The highest value left is the 90^th percentile value i.e. “10 sec” is the 90^th percentile value.

Prerequisite of Load Test Execution

1. Ensure that time slot is booked in calendar on your project name.
2. Check that application is working manually.
3. Check whether the entire server’s are up and in normal condition.
4. If in SiteScope the values against summary of any server are ‘NA’ then restart those servers.
5. If time permits run each script individually for five iterations in VuGen.
6. Refresh the script in Performance Center with the latest version.
7. Check the runtime settings of each script.
8. Check the scenario settings and number of users configured for each script with workload Model.
9. Check for IP Spoofing whether enabled or disabled and change it as per the requirement.
10. Before starting the scenario verify that each script duration, ramp up and ramp down time is correct.
11. Check whether Monitor profiles are configured correctly.
12. Check whether windows profilers are added.
13. Book the same time slot in Performance Center and manually select Load Generators and Controller.
14. Configure number of Load Generator, Controller as per allocated in time slot and take only those Generators and Controller, which are free.
15. When ready to run the test send an alert mail to all related groups.

Key Performance Indicator

Category	Counter	Description	Optimal Value	What if not Optimal	Can Cause
Processor	System\Context Switches/sec	This counter gives the combined rate at which all processors switch from one thread to another. This counter is indicative of hardware devices functioning properly.	Context switches per request (Context Switches/sec divided by Webservice\Total Method Requests/sec) should be low. As a general rule, context switching rates of less than 5,000 per second per processor are not worth worrying about. If context switching rates exceed 15,000 per second per processor, then there is a constraint.	If this counter suddenly starts to increase, it may be an indicating of a malfunctioning device. If CPU utilization is low and very low level of context switching, it could mean that threads are getting blocked.	Increased %Processor Time
Processor	Processor(_Total)\% Processor Time	% Processor Time is the percentage of elapsed time that the processor spends to execute a non-Idle thread.	Less than 85 percent.	This counter suggests either there is too much load on the system or there is some other factor causing the CPU to shoot up. This counter shall be used with % Privileged Time or Processor Queue Length	System Unstability
Processor	System \% Processor Queue Length	% Processor Queue indicates the collection of threads that are ready but not able to be executed because there are threads in execution.	Less than 2	This counter indicates there is too much load on the system or application threading logic needs improvement, If CPU %Time is constantly 85 percent or higher with PQL of 2 or more, then CPU is the bottleneck. If the CPU time is low with PQL of 2 or more, then threading logic shall be investigated.	System Unstability
Disk	LogicalDisk\% Idle Time	% Idle Time reports the percentage of time during the sample interval that the disk was idle.	Greater than 5 %	If the value is less than 5% means disk is really busy. The Value should be greater than 20% typically.	Increased %Processor Time
Disk	LogicalDisk\Average Disk sec/Read	Avg. Disk sec/Read is the average time, in seconds, of a read of data from the disk.	Less than .015 sec	if the value is more than .025 sec that means disk is bottleneck. Response Time would be affected.	Increase Response Time
Disk	LogicalDisk\Average Disk sec/Write	Avg. Disk sec/Write is the average time, in seconds, of a write of data to the disk.	Less than .015 sec	if the value is more than .025 sec that means disk is bottleneck. Response Time would be affected.	Increase Response Time
Memory	Memory\Pool nonpaged Bytes	This counter displays the last observed value of an area in system memory that contains objects that cannot be written to disk because they are still being referenced.	Steady or No more than 10 percent increase since system start up	Increase of this counter indicates a potential memory leak.	System Unstability
Memory	Memory\Available MBytes	This counter gives the amount of physical memory available to processes running on the computer in Megabytes.	25% of physical memory.	It would mean the application is not getting enough memory to operate and could result in application slowness. A gradual decline in Available memory may indicate a memory leak.	Memory Leak
Memory	Memory\Pages/sec	Pages/sec is the rate at which pages are read from or written to disk to resolve hard page faults.		Multiply this value with Physical Disk\ Avg.Disk sec/Transder and if the product of those 2 counters exceeds 0.1, then this would indicatres that RAM shall be added, because paging is taking more than 10 percent of disk access time	Memory Bottleneck
Network	Network Interface\Bytes Total/sec	Indicates the rate when data send and received on the network adapter	Less than 80% of the network bandwidth	This counter will indicate if the Network bandwidth on the Adapter is getting bottleneck	Adapter Bottleneck
Server	Server \Bytes Total/sec	Indicates the rate when data send and received on the network	Less than 50% of the network bandwidth	This counter will indicate if the Network bandwidth is getting bottleneck	Bandwidth Bottleneck