A state-of-the-art server machine is not the end of the matter. Keeping tabs on the system is also a very important aspect of running a Web server. This will unsure that any problems with the machine’s hardware, software or even the physical connection to the Internet, will show up in the analysis. Data gleaned from monitoring, alerts administrators to problems, and can provide positive identification to the source of the problem also.
In a management system, there is a central suite of software that is dedicated to monitoring the Web server and the network. The software can assess the performance of all related objects on the network, e.g. servers, routers or hubs. A terminal is available that allows the administrator to view the results, called the management system. The central suite of software monitors each object, by communicating with each of them through software agents installed on the objects and collecting statistical information. Installing a management system such as this, means a difficult choice has to be made; which of the different protocols, currently available, should be used?
There are basically two categories that network management systems fall into, those which utilise proprietary protocols, and those which comply with a standard. The ISO (Industry Standard Organisation) defines its management system within the CMIP (common management information protocol) system. CMIP might be the official standard, but the most widely accepted standard with the greatest installed base is the SNMP (simple network management protocol) suite. This was defined by the developers of TCP/IP, the Internet protocol. Other systems are also available, such as IBM’s NetView. It has been suggested that CMIP is the most powerful of all the management protocols, and will be widely embraced as the protocol of choice.
The agents running on various objects like the server, firewall, proxy, routers, bridges and hubs can possess an area of memory and a processor, to gather statistical information and pass it on to the management station monitor. The monitor polls each agent and retrieves the statistical information that has been gathered about its own performance. Usually the monitor station can present these results in a graphical format on a GUI-based interface.
Server Performance Tools
Thus far we have seen by and large the management of the entire network. This is of course very important, since your Web server may well be providing Intranet services to the organisation as well as Web services for the outside world. From the Web server perspective, it is more important to monitor the server. It is the task of the administrator to ensure that the server is operating at its optimum performance, ensuring that there are no problems, bottlenecks, or disasters lying in wait. The best way to look out for these is to run server-monitoring tools on the server, to continually check its status.
As mentioned under the section dealing with Networks, NT comes with a good selection of tools. Other operating systems come with a range of similar tools, or tools can be installed from a third part vendor.
Windows NT includes the following utilities:
This utility is a good tool for finding out what is happening around the server at the packet level of network transmission. It monitors all traffic to and from the server, and can be set to filter the information from a particular node or address
This utility provides comprehensive information on how the server is working, and is the main way of detecting upgrading needs. Its main feature is to display information in a chart form, relating to a particular resource on the server. It supports an impressive list of items from different memory objects, CPU and network traffic, and even more Internet specific items like, CGI requests, connections and file throughput.
This utility provides a way of monitoring the resources being used on the server. This includes a view of shared resources, users connected to the server and which is connected to what resource.
This utility provides a graphical view of the CPU and memory usage on the server. It also lists current processes and applications running on the server. These representations allow the administrator to assess the performance of the components. Obviously, if they were being strained, then this would indicate the need for upgrading.