Version 1.1 User's Guide
The HTTP Client: Accessing Remote Web Sites	Previous | Next Contents

The HTTP Source plug-in controls how WebDoubler retrieves requests from the Internet. Configuring the HTTP Client controls how the HTTP source attempts to communicate with remote Web servers and other proxy servers. The WebDoubler window for configuring the HTTP Client is shown below. Using Web-based remote administration to set up the HTTP Client is very similar and includes the same configuration options.

Figure 5: The HTTP Client Plug-In Window

Proxy Pseudonym

The Proxy Pseudonym field is the name of your WebDoubler proxy that is sent to the remote Web servers or upstream proxies. Each time a request passes through WebDoubler, a stamp is added to identify that the request is being transferred via the proxy. The stamp may be used by upstream proxies or the destination Web server for a variety of reasons, or may be ignored entirely. In any case, you should enter a name (or pseudonym) to identify your proxy server, so that upstream servers can identify each intermediate proxy, if necessary.

We recommend selecting a short, simple, descriptive name for each proxy you administer. For example "CompLabProxy", "AccountingDept", and "MaxumProxy" would all be good choices. Using only alphanumeric characters ("a-z", "A-Z", and "0-9") isn't required, but may help to avoid any potential problems with upstream servers parsing the proxy names.

Maximum Connections

The Maximum Connections setting specifies the maximum number of outbound connections that WebDoubler will maintain at any one time. The default setting is 16 connections. In general, the more Web clients on your LAN and the larger your connection to the Internet, the higher you should set the maximum number of simultaneous connections.

Increasing the number of simultaneous connections may improve performance on requests to slow servers or for large download files. For example, if the remote Web server is slow, then connections will take longer to complete, and at the same time use less of your available Internet bandwidth. Since connections will take longer, more simultaneous connections may need to be supported in order to service the same overall number of requests.

On the other hand, decreasing the number of simultaneous connections may decrease the contention for bandwidth on your connection to the Internet, thus providing more bandwidth for active data requests. In other words, setting the number of outbound connections lower can improve performance by restricting the number of connections competing for bandwidth. In this case, some connections may be denied or delayed, but most connections will be processed faster and overall efficiency of the connection increases.

Enable Tunneling (SSL) Protocol

The Tunneling checkbox specifies if tunneling is enabled for SSL connections. When a secure (SSL) connection is established between a Web browser and an SSL server, intermediate computers, including proxies, can not decode the encrypted data being sent back and forth. For this reason, WebDoubler can not process secure transactions as it does other Web connections. However, secure transactions are supported through a process called tunnelling.

In most unencrypted Web connections, WebDoubler understands and even modifies the request being made, so that content filtering, caching, and other actions can be taken. For secure transactions, WebDoubler blindly transfers data between the Web browser and the server, effectively "tunneling" the data without actually understanding it. Enabling SSL Tunneling will allow you to access secure Web servers (and possibly other facilities that use the tunneling protocol) through WebDoubler, but note that secure transactions will not be filtered, cached, or otherwise affected.

Enable Persistent Connections

The Persistent Connections checkbox determines if the HTTP Source should attempt to use persistent (also known as "keep alive") connections to remote servers.

Modern Web servers are capable of maintaining "Persistent", sometimes called "Keep Alive", connections. Without persistent connections, each requested object requires a separate connection to the Web server. This can account for a lot of overhead, since each connection must be opened, processed, and closed independently for every Web page, image, file, etc.

Persistent connections provide the ability to open a single TCP/IP connection, process multiple object requests, and then terminate the connection. This means, for example, that an entire Web page, complete with all of the embedded images, can be processed on a single connection. This is a much more efficient way to transfer multiple objects, and can dramatically improve performance of remote Web sites, especially on slow or very busy Internet connections.

Enable Request Pipelining

Most servers that support persistent connections (described above) also allow for "Request Pipelining". When pipelining is disabled, persistent connections operate by opening a TCP/IP connection to the server, sending the first object request, receiving the first object, sending the second request, receiving the second object, and so on until the connection can be closed.

Pipelined connections operate more efficiently. With pipelining enabled, object requests can be sent before earlier requests have been completely filled. The diagram below shows the difference between pipelined and non-pipelined connections, with grey boxes representing HTTP requests from the client and white boxes representing the responses.

Figure 6: Pipelined vs. Non-Pipelined Requests

Notice that because each response must be received in full before sending the next request, non-pipelined connections are much less efficient. For this reason, you may find it best to enable request pipelining. However, if you discover that one or more Web sites that you access regularly frequently have broken graphic icons or other loading problems, you may want to try disabling request pipelining.

Upstream Proxy

In some situations, it may be desirable (or mandatory) that requests be processed through more than one proxy server. For example, a school might have a proxy server that in turn must go through a district-wide proxy. Large organizations may use proxies in individual buildings or departments, each of which feed up into larger organizational proxy servers. WebDoubler supports "proxy hierarchies" by allowing you to specify "Upstream Proxies" to be used instead of connecting directly with remote Web servers.

If you are using WebDoubler in conjunction with an upstream proxy server, first be sure to check the "Use Upstream Proxy" checkbox.

Proxy Address

The Proxy Address field specifies the TCP/IP address or domain name of the upstream proxy server.

Proxy Port

The Proxy Port field specifies the listening port of the upstream proxy server.

Username And Password

If the upstream proxy server requires client authentication (a username and password), enter the username and password assigned by the upstream proxy manager.

Copyright © 1999 Maxum Development Corporation http://www.maxum.com/	Previous | Next Contents