Hello Everyone,
I am facing lot of concurrency wait events due to "block change tracking
buffer space" when rman L0 backup is run. I see that database gets
overwhelmed with concurrency due to less space for CTWR buffer as shown in
event. I see following problem area . Any recommendations if we should bump
"_bct_public_dba_buffer_size" to higher value ?
NAME
VALUE DESCRIPTION
---------------------------------------------
------------------------------ ------------------------------
-------------------------
_bct_public_dba_buffer_size
0 total size of all public change tracking dba
buffers,
in bytes
NAME
VALUE DESCRIPTION
---------------------------------------------
------------------------------ ------------------------------
-------------------------
_bct_buffer_allocation_max
104857600 maximum size of all change tracking buffer
allocations,
in bytes
SQL> select dba_buffer_count_public*dba_entry_count_public*dba_entry_size
from X$KRCSTAT;
DBA_BUFFER_COUNT_PUBLIC*DBA_ENTRY_COUNT_PUBLIC*DBA_ENTRY_SIZE
-------------------------------------------------------------
25276400
SQL> SELECT * FROM v$sgastat WHERE name like 'CTWR%';
POOL NAME BYTES
------------ -------------------------- ----------
large pool CTWR dba buffer 11386880
snippet from ash and dash is as follows:
%This SECONDS WAIT_CHAIN
------ ---------- ------------------------------
------------------------------------------------------------
------------------------------------------------------------
------------------------------------------------------------
------------------------------------------------------------
------------------------------
12% 14650 -> 1vyayutd1117p:buffer busy waits -> 1vyayutd1117p:block
change tracking buffer space
--
Cheers
Nitin