Page Performance Test Tool page Detailer Pro

Page Detailer Pro is a tool provided by IBM Alphaworks. It is a software used to record browser HTTP requests by inserting probes (Probe) into the client's Windows port stack to obtain data of various types that record browser-initiated HTTP requests.

Figure 1 is a screenshot of a Page detailer record. The light blue callout part is added after, from the above diagram know: Page size. The page size here only contains the size of the resource itself and does not contain the header dimensions of the HTTP headers and other protocol stacks. Of course, Pagedetailer also provides data for total download size. The number of HTTP requests. Individual resource download time. This entire bar represents the total time that a single resource was downloaded, including the time it took to establish a network connection, when the HTTP request was sent, when the HTTP response was received, and when the network was transmitting the entire resource. The time that the browser issues the request header to the response header. Pagedetailer only provides this time, but the fact that the Network listener tool can provide more granular time to record. For example, the HTTP request's sending time, waits for the server response time, accepts the HTTP response head the time. Because in the local area network, accepts the sending head the time to be extremely short, therefore this time can consider is the server response time. In this diagram, adding all the time that the blue bar represents is the time that the server consumes, and the time it takes to remove the overlapping parts is the server response. Note that the time the server consumes is not equal to the time the server responds, and there is a concurrency problem. Network transmission time. Just the time the browser starts to receive the content of the resource to the end of the acceptance. In general, it is the resource size/bandwidth. However, if the server or browser processes the resource in a streaming manner, it also contains the processing time for the browser or server. Browser rendering time. The time interval in any two resource downloads is the rendering time of the browser. Note: Even when a resource is downloaded, the browser may be rendering it. However, these consumption can not be judged by the method of interval time. In general, this part of the time is relatively limited and tends to be ignored. There is no overlap between resource downloads, which means there is no concurrency between these downloads. There is overlap between resource downloads, which means there is concurrency between these downloads.

So how does the concurrency degree count. A simplified method is: concurrency = the sum of individual resource download times/(page download time – Browser rendering time)

The premise of simplification is that the network latency is low in the LAN. The server responds very quickly. Server-side, browser-side no streaming, or quickly

Of course, here is only the introduction of the calculation of the degree of concurrency principle. Based on this principle, it is possible to write programs that read data from Pagedetailer or other network listeners (such as WireShark) to automatically calculate concurrency.