The definitions below explains how the main statistics are measured by NeoLoad.
Count statistics for pages/transactions/requests are the number of full executions. When a page/transaction/request is interrupted (because of error or end of test), then the count number is not incremented.
NeoLoad response time is measured from the time the first byte of a request is sent, until the time the last byte of the response is received, namely including network time to send request and receive response.
It includes DNS resolution time, connection time, SSL handshake time. It does not include think time, time to create the request nor time to wait for an available socket.
When two requests are sent over the same connection, the connection establishment time is only included in the first request response time. As a result, when connection is kept alive through several requests, it is common to have a higher response time on the first request, as it includes connection initialization operated once.
By default, browser parallel connection per host is equal to 6 (can be customised), so it is common to have the 6 first request execution with higher response time.
Web page response time is the time to execute all requests of a page, measured from the time the first byte of the first request is sent, until the time all responses are received. It includes network time to send all requests and receive all responses.
The number of connections available on the population browser settings, and the page playback settings have an impact on the way nested requests are executed, and thus on the Web page response time measured.
The number of simultaneous HTTP1 connections that can be opened on the remote server by one Virtual User can be configured on the population browser settings (see Browser profile). If the Virtual User needs to execute more requests than the number of available connections to a server, then the request execution is paused until a connection becomes available. In NeoLoad, each Virtual User has by default a pool of 6 connections per remote server.
The requests nested under a Web page can be executed either in parallel, or sequentially (see Web pages).
Transaction duration is the time to execute all elements within the transaction. It includes response times of all nested resources (including delays and excluding think times).
Userpath duration is the time to execute all elements within the userpath. It includes the duration of the Init container execution, the sum of the duration of each Actions containers execution, and the duration of the End container execution. Userpath duration includes delays time and excludes think times.
The minimum and maximum response time statistics are measured for each request, page and transaction, and aggregated per time interval.
Length of time interval is defined based on the scenario duration policy. Duration policies with “no limit” or “by iteration” have a fix interval of 5 seconds. Otherwise, time duration policy depends on the duration of the scheduled test.
For tests scheduled to last:
The length of time interval based on the test duration is an advanced setting that can be edited in the file
conf/controller.properties located in the installation directory of NeoLoad.
In section [Compact]:
There is no direct correlation between min/max statistics across requests, pages and transactions:
Let's consider the example below. This example represents 4 User Paths with three pages per run.
Each User Path lasts for 20 seconds. So the min and max of a full User Path equal 20 seconds. However the sum of minimum duration per page is 10 seconds and sum of maximum duration per page is 44 seconds.
Same example works with requests and pages.
The “Time To First Byte” duration is the elapsed time between the time the element started and the time the first byte is received: