These traces show how the libwww HTTP/1.1 robot can perform together with the Apache server when using persistent connections and pipelining. In this example, both the client and the server is a PC running Linux, the server is a big machine but the client is only a 486 with 48M or RAM.
The client walks through the online linux documentation, which consists of approximately 1573 documents with a total of 9.1M of text in just 16.49 seconds (575000.0 bytes/sec) using 1 TCP connection!
If you have xplot, then you can get the traces here and you can also download the raw TCP dumps
If not then you can have a look below. The first is the client to server trace and the second is the server to client trace.