User's Guide
Contents
|
Version 1.1 User's Guide |
| Using WebDoubler |
Previous |
Next Contents |
Once you have started WebDoubler, made any configuration changes necessary (including changing the administration username and password), and configured your Web browser(s), all Web surfing will be done through WebDoubler. Using WebDoubler is transparent to users surfing through local Web browsers, but there are a number of tools provided within WebDoubler itself to help you monitor Web access and WebDoubler itself.
When WebDoubler is first launched, the Statistics window is automatically opened. Thereafter, the window is opened by going to the "Server" menu and choosing "Statistics". The display includes a number of general information items reflecting basic operation of the server. Please note that all statistics are reset each time WebDoubler is restarted.
"Client Requests" represents the total number of requests clients using WebDoubler have made. Remember that each object requested by a client represents a request, including HTML pages, graphics, etc. Therefore, the typical Web page (which includes a number of embedded images) may account for several Client Requests.
"Bytes Served" indicates the total number of bytes sent in response to client requests. This includes both HTTP headers and the actual content of each object served.
"Free Memory" shows the amount of available memory in WebDoubler's application space. It is important to note that most of the key functions of WebDoubler pre-allocate memory at startup time. When running WebDoubler with only the plug-ins that came in the original package, it is unlikely that available memory will become a problem (unless you set WebDoubler's application partition lower than the required setting). However, as you add new plug-ins, you will want to verify that the available Free Memory remains at a reasonable level. See the documentation that accompanies additional plug-ins, and the WebDoubler page on the Maxum Web site, for additional information.
The volume of traffic WebDoubler can handle will vary with your network speed, the speed of the CPU WebDoubler is running on, hard disk speed, and even the configuration you have defined for WebDoubler. For additional information on performance tuning and recommended CPU configurations based on peak loads, visit the WebDoubler page on the Maxum Web site.
"Simultaneous Connects" shows the highest number of client requests that have been active at a single time. This number is limited by the "Max. Connections" setting on the WebDoubler Configuration window. After you have been running WebDoubler in a production environment for a few days or a week, you may want to reset the Max. Connections value based on the actual number of Simultaneous Connections experienced on your server. In general, Max. Connections can be set to be slightly larger than the peak number of Simultaneous Connections. This will ensure that there will always be enough available connections to process client requests while minimizing the memory and CPU demands required to support each available connection.
If Simultaneous Connections quickly reaches the Max. Connections value, you may want to increase the maximum number of connections allowed. Keep in mind, however, your Internet link is only capable of a certain number of connections before performance is reduced to unacceptable levels. By supporting more simultaneous connections, you may slow down the performance of all connections, especially when your link to the Internet is slow. Also, WebDoubler will queue incoming requests when no connections are available. So even when the Peak Simultaneous Connections reaches the defined Maximum Connections, all requests will be responded to correctly (after a delay).
The setting for Max. Connections is also related to the maximum number of outbound connections allowed by the HTTP Client plug-in. See the section below on statistics for Outbound Requests and the documentation for the HTTP Client plug-in for more information.
These values give a good indication of the performance improvement and increased bandwidth utilization realized by running WebDoubler. Requests served from cache are not only served very quickly, but they do not have to be retrieved from remote servers over your Internet connection. This has the compound effect of improving performance of requests that are served by accessing remote servers, because more bandwidth is available on the Internet connection.
Reviewing the Cache Performance numbers can be helpful in optimizing the cache "aggressiveness" setting. See the Cache Manager documentation for additional details on this and other cache settings.
"Peak Connections" is the maximum number of active connections between WebDoubler and remote Web servers at a single time. Peak Connections is very similar to Simultaneous Connections, described above, except that it refers to the number of outbound requests being made by WebDoubler instead of the number of inbound requests being made by LAN clients. The value can be used to fine tune the HTTP Client "Maximum Connections" setting, just as Simultaneous Connections can be used to determine the servers Max. Connections.
"Delayed Connections" represents the number of times the HTTP Client has attempted to retrieve an object from a remote Web server but has been delayed because no outbound connections were available. In these cases, the HTTP Client was forced to wait until another connection completed its request before being able to initiate the request with the remote server. In many cases, this delay should have been very short. However, if the number of Delayed Connections grows rapidly at your installation, you may need to increase the number of Maximum Connections allowed to the HTTP Client plug-in.
Limiting the number of outbound connections may actually be beneficial in some situations. By limiting the number of outbound connections, each will be able to use a greater fraction of the bandwidth of the connection to the Internet, and should complete more quickly. This is especially true when your Internet connection is relatively slow and users are making requests from Web sites that respond quickly.
The Activity Monitor displays messages that show exactly what WebDoubler is doing. The WebDoubler application and any and all plug-ins may send messages detailing problems, warnings, or basic information about how requests are being processed.
Each line in the monitor displays a single message in three columns. The first columns shows the connection number for the request. All of the messages related to a particular request will have the same connection number. This is an important piece of information, since a large number of requests may be processed at a single time. Note that connection numbers are not sequential, but are recycled as requests are completed and new requests are received. The second column shows the time each message was received, and the last column shows the message itself.
You can select the message level you would like to see in the monitor window. This allows you to filter messages so that you see exactly the level of detail you need. The available message levels are:
| Debug | Extremely detailed messages describing most low-level functions |
| Verbose | Detailed messages describing most high-level functions |
| Basic | Simple messages that note processed requests |
| Warnings | Messages are sent only when unexpected processing is required |
| Error | Messages are sent only when errors occur |
| Suspend | No messages are sent to the Activity Monitor at all |
An increasing amount of CPU resources are required to animate the Activity Monitor as the message level is increased. During normal processing you will want to set the message level to "Basic", "Warnings", "Error", or "Suspend" to avoid unnecessary use of CPU resources recording and displaying messages.
If you would like to record the messages displayed in the Activity Monitor window, select an Error File and Error Level in the Server tab of the WebDoubler Configuration window. Note that the error log and the activity monitor operate independently, so you may save the messages to a file without displaying them in the Activity Monitor window, or vice-versa.
Real time graphs are available to review current and past proxy traffic patterns. Note that these graphs are not intended to provide detailed activity analysis. For this, WebDoubler includes an activity log function that can maintain a Common Log Format (CLF) log file for processing by most popular log analyzers. The graph window is shown below.
Four real time graphs are available. The first three, "2 Minute", "2 Hour" and "5 Day" graph the number of connections completed over the past two minutes, two hours, or five days, accordingly. The display graphs this activity relatively, so the height of the bars does not correspond to any preset number of connections. While the actual number of connections can't be determined from the graphs, general activity and peak loads are represented.
The fourth graph, labeled "Threads", is a display showing current connection activity. The meaning of each bar in the graph is fairly technical, but in general the graph can be used to assess the level of activity of the server at any given time. When the graph is active, with many bars displaying fluctuation, the server is busy. When the graph is static, with many threads showing no activity, the server is idle.
You will notice that the first bar (or thread) is red. This bar represents the "main application thread", which is responsible for managing the user interface, receiving events from the operating system, and performing other routine Mac OS application tasks. The next bar is yellow, and represents the "listener". This thread is responsible for "listening" to the network for incoming client requests, and then assigning them to connection threads.
The block of blue threads are the connection threads that process client requests from the browsers on the LAN. These are sometimes called "workers", each being responsible for processing a single client request. Once a client request has been completely processed, the blue connection thread will settle back to inactivity and wait to be assigned another connection to work on.
Finally, there will usually (but not necessarily) be one or more green threads displayed. These are threads that are dynamically created to perform some task. For example, the HTTP Client plug-in creates a green dynamic thread to monitor and close persistent connections to HTTP/1.1 compliant servers. The cache manager also dynamically creates a number of green threads to perform various functions.
You will notice that the graph will change as green dynamic threads are created, perform their task, and are destroyed. Also, as with the other activity graphs, bar height is relative to the activity of other threads and there is no set activity level that correlates to any particular bar height.
| Copyright © 1999 Maxum Development Corporation http://www.maxum.com/ |
Previous |
Next Contents |