News Database

Data cached by the News Cache have to be stored in a database. Hence the News Cache has to maintain a news database consisting of the newsgroup, article and overview database similar to the database of the news server.

Newsgroup and Article Database

  Standard news servers like INN [Sal92] and CNews [CS] use the local file system to store their newsgroups and articles. Each newsgroup uses its own directory and the articles are stored using a file for each article. To provide a unique file name for each article, the article number is being used for the file name. This straightforward approach has several disadvantages:

• News articles are usually very small (about 1-4 Kilobytes). Disk space is allocated in blocks which are typically between 512 and 4096 Bytes. Thus disk space is wasted depending on the block size (approximately between 10% and 25%, as shown in Figure 4.2).
• The total number of files is limited in most file systems. Thus free disk space may be available, but cannot be allocated if the maximum number of files has already been reached.

  

Abbildung 4.2: File system space overhead

Hence, a new and better approach should be found for the News Cache and future news servers. An alternative approach is the use of a database for each newsgroup, where all the newsgroup's articles are stored. However, this requires the availability of a powerful database. Nevertheless we adopted this approach. On the one hand the News Cache does not store as much articles as the news server does and on the other hand we did not have the resources and time to design a new article database.

Overview Database

The overview database provides a short summary of the articles available within a newsgroup. The news server can compute this data when the clients requests it since all articles are stored within the news server's database.

The News Cache does not necessarily store all the articles of a newsgroup. Therefore the News Cache cannot compute this database on the fly using the article database. Hence, we will have to use another database storing the overview informations for each article. The design and implementation of this database will be presented in section 5.5.3 and section 6.5.3, respectively.