The purpose of this article is to provide a basic definition for both HTTP Client Workers and HTTP Concurrent Connections as well as provide some documentation related to both statistics.

• HTTP Client Workers are processes the proxy starts when it receives HTTP requests through one of the available HTTP Methods (GET, POST, CONNECT, HEAD, OPTIONS, etc). After the proxy replies to an HTTP request with any kind of response code, the HTTP Worker ceases to exist.

• HTTP Concurrent Connections are the number of TCP connections the proxy is receiving in its HTTP/FTP or Tunneled Proxy Services at a given time (usually “Explicit HTTP” for Explicit or “External HTTP” for Transparent by default).

Usually, when a client connection begins (after the TCP Handshake), the client will send an HTTP request (normally a GET request). This will create a new HTTP worker until a response is received. After that, the client can send more HTTP requests within the same TCP connection, so new HTTP workers will be created and killed for each one of those requests.

There can be connections without HTTP workers, but in order to have a worker, we require an existing connection. For this reason, the number of concurrent connections is usually higher than the number of HTTP workers.

Each ProxySG model has its own HTTP concurrent connections and HTTP workers’ limit. It’s important to note that this limit applies to both statistics, so both of them should always be checked when troubleshooting performance issues.