Input/Output operations are one of the major factors which affect the performance of any system. In case of Oracle database it may reduce the response time and regularity of work. This I/O operation is considered very expensive because the reading function has to wait until the physical Input/Output operation to complete. The disk operations are considerable slow than a row’s access in the data buffers in an Oracle database (approximately row access is 1000 times faster than I/O operations). Therefore anything that can minimize the disk access or reduce the bottleneck caused by the informalities for files on disk can considerably improve performance.
For improving the disk operations Oracle has put forward some techniques namely Hash Tables Clusters and Oracle Row Re-sequencing. The hash tables makes uses of the hash table techniques which order the data in an easily accessible way and uses the hash key functionality. The row re-sequencing reorder the row is more precise way such that the amount of time that would be taken to access the row is minimized. These techniques have a lot of inside functionality that is quiet advance and more referencing is needed to totally understand them.
The queries which access a set of rows from an unordered table take huge amount of Input/Output time since the index would separate each data block and the query engine have to search in each of the data block separately. But by using 10g hash table clusters, single table clusters or manual row re-sequencing methods the fetch time can be reduced considerably. The performance can be measured by the clusthering_factor in dba_indexes. The main reason for the reduction in I/O is because the rows are divided into several data blocks by using the various techniques that is mentioned above and only a single block is needed to be accessed in order to get the desired results.
In online transaction environments the data is accessed based on the primary indexes. In such systems the row re-sequencing and done is such a way that the rows are ordered in contiguous blocks by following the same order as the primary index and this can actually reduce the physical I/O and improve response time. This method is only application when the fetch is for multiple records and when the Oracle used Index range scans. The ordinary primary –key unique access don’t actually benefit from row re-sequencing.
These methods that are mentioned above are grouped under the clustering techniques that had been introduced by oracle. The hash tables had evolved considerable each time a new version is introduced by oracle. There are lot of chances and new methods that are to be implemented in the days to come.