SOS Oracle Detail Latches

The Oracle Detail Latches Screen

The contention for various buffers in SGA is solved using latches. The SOS Oracle Detail Latches screen presents latches statistics.

To access the Oracle Detail Latches screen from any SOS display screen:

Type s from the SOS Enter command: prompt to view the Screen Selection Menu.

From the Screen Selection Menu, enter L (upper case). An example of the screen is shown in Figure 35.1.

Figure 35.1 SOS Oracle Detail Latches screen

Oracle Detail Latches Display Items

The data values in each section of the Oracle Detail Latches screen are described in the next table. Unique data items are described in the following sections of this chapter.

Table 35.1 SOS Oracle Detail Latches data items

Data Item	Description
Gets	The number of successful gets.
Misses	The number of successful misses.
Im Gets	The number of successful immediate gets. A get is considered to be immediate if the requesting process specifies that it does not wish to wait for a latch on its data to be released.
Im Misses	The number of misses for immediate get requests.

LRU Latches

The LRU section has two sub-sections of data.

Table 35.2 SOS Oracle Detail Latches data: LRU Latches

Data Item	Description
Cache Buffer Chains	This latch is required when user processes try to scan the buffer cache from SGA. Adjusting DB_BLOCK_BUFFERS can reduce the contention for this latch.
Cache Buffer LRU Chain	This latch is required when the user processes try to scan the LRU chain that contains the dirty buffers from the buffer cache. Increasing DB_BLOCK_BUFFERS and DB_BLOCK_WRITE_BATCH can reduce the contention for this latch.

Redo Log Buffer Latches

An Oracle process has to obtain redo copy latch before the redo allocation latch. After it obtains both, it makes the allocation, and after that, it frees the redo allocation latch. Only after it makes the copy, it will free the redo copy latch. So the redo copy latch is kept more than the redo allocation latch.

You can reduce the conflict for this latch by reducing the copy time. To do that, you could reduce the LOG_SMALL_ENTRY_MAX_SIZE parameter. Usually the latch contention occurs on multi- processor machines. The LOG_SIMULTANEOUS_COPIES parameter, which determines the number of redo copy latch mechanisms, with the default value equal to the number of processors, should be increased if contention occurs.