RE: Latch Contention
I have a threaded batch process running in which all
the threads are pretty much doing the same thing. I have another report
which shows total IO, CPU, REDO and more by session. When you look at this
report (basically workload) all of the #'s and times are pretty much the
same. Meaning all of the threads are processing about the same amount of
work. However, if you look below I get these usual spikes in Buffer Busy
waits and they correlate to specific sessions. it looks like the spikes
occur during a checkpoints. It is strange because it looks like just a
couple session accumulate all of the buffer busy time instead of all of the
sessions.
Total Latch
Buffer Db File Db File D Path D
Path Log Buf Log File
SQL*Net
SID Seconds Free Busy Sct
Read Seq Read Read Write Enqueue
Space Sync to Cli
----- ------- ------- -------
-------- -------- ------- ------- ------- ------- --------
-------
10
0 0
0
0
0 0
0 0
0
0 0
8 0
0
0
0
0 0
0 0
0
0 0
9 1
0
0
1
1 0
0 0
0
0 0
41 109
4 15
3 83
0 0
0
0
4 0
19 114
2 0
24 44
0 0
0 0
44 0
31 125
4 34
3 80
0 0
0
0
4 0
25 126
3 17
3 98
0 0
0
0
5 0
27 127
4 10
3 106
0 0
1
0
4 0
29 151
4 31
2 111
0 0
0
0
3 0
38
1784 4
26 3
1747 0
0 0
0
3 0
40
1788 4
28 1
1751 0
0 0
0
4 0
37
1941 3
1807 2
125 0
0 0
0
4 0
32
2053 3
19 3
2024 0
0 0
0
4 0
28
3578 3
1874 3
1693 0
0 0
0
4 0
34
3798 4
13 3
3775 0
0 0
0
4 0
14
4266 4
33 2
4223 0
0 0
0
3 0
24
4892 3
9 2
4875 0
0 0
0
4 0
26
5090 4
21 2
5058 0
0 1
0
4 0
39
5371 5
75 2
5287 0
0 0
0
2 0
35
6737 2
13 2
6717 0
0 0
0
4 0
23
7075 5
2525 2
4539 0
0 0
0
3 0
33
9435 3
24 2
9401 0
0 0
0
3 0
30
9835 3
24 3
9802 0
0 0
0
3 0
36
10481 4
32 2978
7463 0
0 0
0
4 0
