Friday, August 13, 2010

Data Guard optimizing ASYNC redo transport by log buffer

Data Guard redo configuration asynchronous mode, the Log Network Server (LNS) process will attemp to read redo directly from the log buffer. If the redo to be sent is not found in the log buffer, then the LNS process will go to the Online Redo Log (ORL) to retrieve it. Size the log buffer so that LNS is always able to find the redo that is needs to send within the log buffer.
The log buffer hit ratio is tracked in the view x$logbuf_readhist. A low hit ratio indicates that the LNS is frequently reading from the ORL instead of the log buffer. If there are periods when redo transport is coming close, but is not quite keeping pace with your redo generation rate, consider increasing the log buffer size in Data Guard 11g to achieve a favorable hit ratio.

Increase the log buffers
improves the read speed of the LNS process.

DESC x$logbuf_readhist

Name                Null  Type
------------------- ------ ------------
ADDR                       RAW(8)
INDX                       NUMBER
INST_ID                    NUMBER
BUFSIZE                    VARCHAR2(20)
RDMEMBLKS                  VARCHAR2(24)
RDDISKBLKS                 VARCHAR2(24)
HITRATE                    NUMBER
BUFINFO                    VARCHAR2(20)
select bufsize, rdmemblks, rddiskblks, hitrate from x$logbuf_readhist;

------- --------- ---------- -------
  2665K 484903388   42044064      92
  3197K 506308671   20638781      96
  3730K 508354202   18593250      96
  4263K 512373070   14574382      97
  9059K 523792275    3155177      99
  9592K 523808659    3138793      99
 10125K 523818899    3128553      99
 10658K 523822995    3124457      99