ARP means Address Resolution Protocol and it is the standard method of finding the host’s hardware address when only the network layer is known. It is thus a critical part of IP networking. Typically, in a network, every resolution will involve the exchange of messages between devices. Each time an ARP message is sent by a device, this ties up the network by consuming some bandwidth. Although ARP messages are not large, sending messages for every IP datagram would overwhelm the network. This might also result into some considerable time wasted, especially when compared with simpler direct mapping resolutions. Moreover, each ARP message is broadcast, and every device on the network will spare some CPU time to examine the content of those messages. The solution to this problem which would have grave implications for the network operability, is the use of caching. The effectiveness of caching can be demonstrated in the following way. If you send a request to an online site on your computer, the request will need to go through the network’s local router. You may later want to go back to the same site and click on a link to it. The request will also go through the router. Everything done in that network will probably be done through that router or Default Gateway. If the router has to do the same resolution over and over again, even for requests that had been made and processed earlier, that would be a waste of time. It’s almost akin to consulting a directory every time you want to call your friend of 20 years.

Caching envisions a situation whereby after a device network layer has been resolved to a data linked layer address, that link between the two is saved in the memory of the device for some time. When the address is needed sometime later, the device quickly looks it up in its cache. This means that instead of doing a broadcast every time there is a request, the broadcast is only done once.

Thus, an ARP cache can be described as a table which contains matched sets of the hardware and Internet Protocol (IP) addresses. Each device on a given network has its own cache. There are two ways in which entries ca be kept in a cache for quick resolution. These are:

* Static ARP Cache entries. Here, the address resolutions have to be manually added to the cache table and are retained on a permanent basis.

* Dynamic AP Cache entries. This is where the Hardware/IP addresses are kept in the cache by the software as a result of successfully completing past resolutions. The addresses are kept on a temporary basis and are then removed afterwards.

ARP Cache transforms what would otherwise be a wasteful process into one where time is used efficiently. It might, however encounter some problems. The cache table must be maintained. There is also the possibility that the cache entries might become stale with time, so the settings should be that the cache entries must expire after certain period of time.

For more information read What is ARP Cache?

http://www.tech-faq.com/arp-cache.shtml

http://www.symatech.net/arp-cache