Log isolation for linux File Systems

Log isolation in linux File System 1. The ext3 File System adds the log function on the basis of ext2 to ensure data consistency like a database. Ext2: Write block before, and then write inode! Ext3: added a log area (conducive to data consistency). Write inode first! After the data is written to the file system, clear the log. however, large data such as oracle and mysql is based on log files and can be separated to ensure data consistency and efficiency. Ext3 file systems are embedded by default, and small servers are not considered. For large servers, disk I/0 is slow, and data writing and reading efficiency is directly affected, therefore, the data zone and the log zone are isolated to reduce the I/O burden and speed up Data Writing efficiency. At the same time, the following steps are provided to facilitate the recovery of the log zone isolation: 1> unmount the partition, make sure that the partition does not use 2> dumpe2fs/dev/sda1 | egrep-I (journal | size) view information 3> tune2fs-O ^ has_journal/dev/sda1 remove internal default log functions 4> mke2fs-O journal_dev-B 1024/dev/sda5 format a Log Device 5> tune2fs- j-J device =/dev/sda5/dev/sda1 Add the log device to the partition 3. you can also specify the frequency at which logs are synchronized to the file system: mount-o commit = 15 interval at which logs are synchronized to files, which may result in data loss. working modes of log file systems: different working modes have different efficiency, and different data security data = ordered by default, only metadata is written into the log area data = journal data, metadata is written to the log area data = writeback. Only metadata is written, but the submission order is not guaranteed.