| Name | Description/comments | param 1 | param 2 | param 3 |
| db file scattered read | read multiple (up to the value of db_file_multiblock_read_count) db blocks as part of a table or index full scan into the buffer cache. Because multiple blocks are read, the individual blocks must usually be «scattered» to different regions in the buffer cache, hence, the more descriptive name is db multiblock read | file number | block number | number of blocks |
| db file sequential read | read one block into the buffer cache when reading an index block or a data block identified by a rowid. The database reads the blocks in a sequence, one here, one there, hence db file sequential read. The more descriptive name is db single block read | file number | block number | number of blocks (should always be 1). See also MOS note 34559.1. |
| db file parallel read | Oracle performs (during recovery operations or buffer prefetching optimization) in parallel reads from multiple datafiles to non-contiguous buffers in memory (PGA or Buffer Cache). The more descriptive name is db list of block reads. | files | blocks | requests |
| db file single write | | file number | | |
| buffer busy waits | Multiple processes try to access the same buffer in the buffer cache. See also MOS note 34405.1 | | | |
| direct path read | Read blocks directly into PGA (as opposed to reading them into the buffer cache). See also MOS notes 50415.1, 1476089.1 and 1475655.1 | file number | first dba | number of blocks |
| direct path read (lob) | Special multi-block I/O's (using temporary tablespaces for sorting because memory is exhausted, parallel execution, readahead for buffer prefetching, direct load, LOB related I/O | | | |
| free buffer waits | Waiting for free space to become available in the buffer cache to read a data block. This wait causes DBWR to write dirty blocks to their data file. See also MOS Note 62172.1 | | | |
| log buffer space | waiting for the LGWR process to make room in the redo log buffer. The more descriptive name is log buffer full - LGWR bottleneck | | | |
| PGA memory operation | Memory allocation or deallocation | number of bytes(?) | ? (always 1?) | ? |
| log file switch (checkpoint incomplete) | Blocks a commit. See also MOS Notes 34592.1 and 1376916.1 | | | |
| log file switch (archiving needed) | Blocks a commit. | | | |
| log file sync | When commiting a transaction, this event waits until LGWR has written the corresponding redo information to the redo log files. The more descriptive name is commit: log file sync | | | |
| SQL*Net message from client | This event is expected if the client is sitting idle. In contrast, if the client is fetching a large resultset with many records, the time waited in this event might be significantly decreased by increasing the fetch size. | driver id (the 4 integer bytes store the ASCII value of a 4-character string) | number of bytes | |