6.8.1 Controlling communication by Persistent Connection
The persistent connection is functionality used for connecting the TCP connection established between the Web client and the in-process HTTP server and keep using the TCP connection between multiple HTTP requests. By using the persistent connection, the time required for establishing a connection between the Web client and Web server is reduced and an attempt is made to reduce the processing time and lessen the communication traffic.
In the in-process HTTP server, you set up the following items to control communication by the Persistent Connection:
-
Upper limit for the number of persistent connections
By setting the upper limit for the number of persistent connections, you control the number of Web clients that can continuously process requests with one TCP connection. If the number of TCP connections exceeds the specified upper limit, the connection disconnects after the processing of the request ends. As a result, you can secure the threads for processing the new requests and prevent the request processing threads from being used exclusively by a specific client.
-
Upper limit for the request processing frequency of persistent connection
By setting the maximum value for the request processing frequency of the persistent connection, you can control the processing when there are continuous requests from the same Web client.
When the request processing frequency of the persistent connection exceeds the specified upper limit, the connection disconnects after the processing of the request ends. As a result, you can prevent the request processing threads from being used exclusively by a specific client.
-
Persistent connection timeout
By setting a timeout for the request waiting time of the persistent connection, you control the request waiting time of the persistent connection. If there is no request to process requests until the specified timeout period expires, the TCP connection disconnects. As a result, you can prevent the TCP connection from being continuously occupied in an unused state. Also, even if you specify 0 for the request waiting time of Persistent Connection so that a timeout does not occur, if the number of requests exceeds the upper limit on the number of requests that can be processed, the connection is disconnected.
Note that a disconnected Web client will try to connect and send requests again.