There are many different types of servers in the computer world. There are network servers, application servers, database servers, email servers, etc. depending on the setup and usage, true server computers may be used as servers. In small to medium sized networks, a desktop computer may be configured as a server.
When an ordinary desktop computer is setup as a server, its hardware is that of an ordinary computer. Its usage is that of a server and no other work is setup on that machine. Therefore, in such a scenario, there is no difference in the memory of the machine. The only difference that it may have from the other machines on the network is that it may have extra memory, a higher speed processor and a bigger hard disk drive.
In very large computer setups, there are server machines, which are different from the ordinary desktop. The difference is that they may have more than a single processor. They may have two or more processors. The hard disks of a server machine maybe SCSI hot swappable disks, or fixed hard disks. As the processors are extremely fast, as are the hard disks, the memory in servers also needs to have a high speed.
The memory in a server is expected to be fast, and is called ECC memory. ECC stands for error correcting code memory. This memory system tests and corrects any errors in memory without the processor or user being aware of it. The error correcting code generates a checksum when data is loaded in memory, and when it unloads, the checksum is recomputed and if an error is detected it is automatically corrected. This ensures that the data passed in the server is correct.
All desktop, laptops, and tower computers use different types of memory classed on their speed and structure. They are known as SDR, DDR, DDR2, etc, and as the personal computer world has seen the personal computers getting faster and faster, the memory speeds have also been enhanced over time.
If you put a high-speed memory in a slow speed processor machine, it will probably burn out and the same rule applies vice versa. The memory speed and the processor speed have to be synchronised to work properly.
In desktop computers, you may have had incidents of a memory chip getting defective, or the computer gives a memory-addressing fault. This never happens in a server machine. Server machines are supposed to be zero-fault tolerance machines. The applications that are supposed to perform cannot afford any downtime.
High-end servers have a very robust architecture, as they are used to perform very specific jobs. Therefore, they are built such that the possibility of a fault occurring is minimised. In case, a fault does occur, the machine will not crash or the application will not stop. Therefore, a proper server machine does have different memory from a non-server machine. Very few people are aware of this, as the environments in which they work don’t use high-end servers.
Jeremy Sandcastle is a technology consultant. Learn more about Cisco Approved Memory