Impact of Oracle row migration

Row migration cannot be avoided. Increasing the size of each block can reduce the number of possible rows for row migration, but it also results in a waste of space. This balance point needs to be determined based on different applications. & Ldquo; row Migration & rdquo;, Oracle Data is saved by block. If a data disk space cannot save a certain data (for example, if it was 1 K before, update to 2 K now, if the free space of the current block is less than 1 K), the new data will be saved to another new block, and a new address connection will be saved in the previous block.

Row migration cannot be avoided. Increasing the size of each block can reduce the number of possible rows for row migration, but it also results in a waste of space. This balance point needs to be determined based on different applications.

For "Row migration", Oracle Data is saved in blocks. If a data disk space cannot save a certain amount of data (for example, if it was previously 1 K, it is now updated to 2 K, if the free space of the current block is less than 1 K), the new data will be saved to another new block, and a new address connection will be saved in the previous block.

For example

Data is stored in one block, with two free spaces in the middle. Now the fourth row of data needs to be resized. If the free space of a single block is insufficient, check whether the merged free space meets the requirements.

If the free space after evaluation does not meet the requirements of the space, a "Row migration" occurs"

Store the fourth row of data in a new block, and then store the connection of the new address in the original block. This is the process of row migration.

What is the impact of migration in line 2?

If you read this row through an index, the index points to the original block, and the block points to the new block. To obtain specific row data, you can obtain row data instead of executing two or more I/O statements. This is not a big problem, or even cannot be noticed. However, if the proportion of such rows is large and a large number of users access these rows, you will notice this side effect. The speed of accessing this data starts to slow down (Additional I/O and I/O-related latencies will increase the access time), and the buffer cache efficiency starts to decrease (two blocks need to be cached, if the row is not migrated, only one block needs to be cached. In addition, the table size and complexity increase. For these reasons, you may not want to migrate rows.