3 replies
by alpha-male
Here is what I found from Librenix:
The most obvious one is that a client has to deal with only one user at a time, while a server has to handle multiple clients at the same time. This means that a machine, being deployed as a server, has to takea much higher load than a client.
Also, unlike a client PC that has to deal with bursts of activity (and load) on its sub0systms (the hard disks is not always pumping out data, nor is the processor always crunching number in a PC); the server isfaced with sustained load on its sub-systems because multiple users access the machine and use its resources. Its hard disk maybe in the midst of serving data to one user, when it receives a request for data from another. So, it has to churn out data for sustained periods of time.
At the lowest end, it’s possible to plus in a workstation into your network and use it for small workgroups for file and print, or even email or proxy services. Otherwise, you need to have a machine specifically designed as a server.
The need to handle sustained loads, therefore, defines the basis of server design. Thus, a server would have more memory, hard disks that can five better throughputs (SCSI hard disks, for instance), RAID to handle the higher data crunching requirements, and even a different chipset to control all of them. Servers, normally, also have a greater number of PCI slots for add-on cards.
The cost to the organization of having a server shutdown is high, as it can affect all PCs in the office. To avoid such a situation, critical parameters of the server’s performance need to be continuously monitored and if possible, managed. Server have additional hardware (and software) to enable this hands-off monitoring and management. Some of these may, depending on teh vendor and the model, be built on the motherboard itself, or may be available as optional add-on cards.
Hope this helps! Good luck
This discussion is closed to new replies.
Trending Networks posts
Sort by Trending