As pointed out in section 3.4.3 a News Cache can be used in areas with an unreliable or slow Internet link (Figure 4.1(a)). In this case the News Cache reduces the transferred data to the absolute minimum. Only those articles being requested by a news client will be transferred from the feeding news server. Articles will be retrieved from the news server only once, since articles that are requested repeatedly are already stored in the News Cache's news database.
The reduction of the network bandwidth can be calculated using the following formula, where ni is the number of accesses and szi the size of article i.
![\begin{displaymath}
Gain=\sum{(n_i-1)*sz_i} \quad \mbox{[Bytes]} \end{displaymath}](img15.gif){kind=small}
A News Cache can replace a leaf node news server since a News Cache requires less hardware and less network bandwidth. Figure 4.1(b) shows a scenario, where a leaf node news server is replaced by a News Cache.
The News Cache reduces the required network bandwidth between the news cache (the former leaf node news server) and its feeding news server. The news cache also reduces the hardware requirements, because it uses less disk space and less CPU time than a news server (the hardware requirements necessary for a news server have been explained in chapter 3 in detail.). To use fewer hard disk space the News Cache has to expire older articles. In case an expired article will be requested again, it has to be requested from the news server again.
The News Cache can be used to multiplex between different news servers. This can be necessary if a user needs to access newsgroups from both news servers (e.g., because one server holds the local news groups and the other the globally accessible groups) and his news reader is not able to distinguish between different news servers. In this situation the user connects his news reader to the News Cache instead of connecting it to the news server directly.
The News Cache can also reduce the load of heavily used news servers (Figure 4.1(d)). This reduces the problem to catch up a backlog. (see section 3.1 for an explanation of this problem). Instead of directly connecting to the news server the user's news reader connects to one of the news caches. Users may connect to whatever cache they like, because the contained articles have the same article numbering as the news server itself.
However, the cache server cannot replace non leaf node news server. These are news servers with links to several other news servers. Non leaf node news servers collect data from many news servers and have to store all arriving articles for distribution. Since the News Cache does not include this functionality, it is impossible to replace these news servers by a News Cache.
![\begin{figure}
\centering
\subfigure[Slow link to news server]{
\epsfig {file...
...eps/nntpcache_nsload.eps}
}
\epsfig {file=eps/nntpcache_key.eps}\end{figure}](img16.gif)