Versions Compared

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

The idea is that we use jCoreDB's FileSystem and PageBuffer to realize the index. The parameters should differ. So for instance smaller segment sizes and page sizes may be required. Data is stored in containers, containers are divided in segments and segments are made of blocks (pages). So an i6ndex i s index is represented by a container. All we need to specify is the number of levels. The next level then contains twice as much elements as the previous one (factor 2). So the first level is realized by using one segments, the second one by using 2 segments, the third one by using 4 segments and so on. The segment size is not constant. Instead it depends on the number of elements those should be stored. The last level can be extended by new segments in order to have room for additional data. An extend happens only on the last level. It is necessary to insert new data by keeping the existing order of the values.

...