原来是UNDO表空间。那么另一个疑问就会来了,为什么在UNDO上产生了如此之多的单块读?首先要肯定的是,这条简单的查询语句,是进行的一致性读。那么在进行一致性读的过程中,会有两个动作会涉及到读UNDO块,延迟块清除和构建CR块。下面我们用另一个脚本来查看会话当时的状况:
SQL> @snapper all,out 5 3 4038
Sampling SID 4038 with interval 5 seconds, taking 3 snapshots...
setting stats to all due option = all
-- Session Snapper v3.52 by Tanel Poder @ E2SN ( http://tech.e2sn.com )
-------------------------------------------------------------------------------------------------------------------------------------
SID, USERNAME , TYPE, STATISTIC , HDELTA, HDELTA/SEC, %TIME, GRAPH
-------------------------------------------------------------------------------------------------------------------------------------
4038, BILL_MY , STAT, session logical reads , 488, 97.6,
4038, BILL_MY , STAT, user I/O wait time , 429, 85.8,
4038, BILL_MY , STAT, non - idle wait time , 429, 85.8,
4038, BILL_MY , STAT, non - idle wait count , 377, 75.4,
4038, BILL_MY , STAT, physical read total IO requests , 377, 75.4,
4038, BILL_MY , STAT, physical read total bytes , 3.13M, 625.87k,
4038, BILL_MY , STAT, cell physical IO interconnect bytes , 3.13M, 625.87k,
4038, BILL_MY , STAT, consistent gets , 488, 97.6,
4038, BILL_MY , STAT, consistent gets from cache , 488, 97.6,
4038, BILL_MY , STAT, consistent gets from cache (fastpath) , 8, 1.6,
4038, BILL_MY , STAT, consistent gets - examination , 478, 95.6,
4038, BILL_MY , STAT, logical read bytes from cache , 4M, 799.54k,
4038, BILL_MY , STAT, physical reads , 382, 76.4
4038, BILL_MY , STAT, physical reads cache , 382, 76.4,
4038, BILL_MY , STAT, physical read IO requests , 377, 75.4,
4038, BILL_MY , STAT, physical read bytes , 3.13M, 625.87k,
4038, BILL_MY , STAT, db block changes , 9, 1.8,
4038, BILL_MY , STAT, consistent changes , 469, 93.8,
4038, BILL_MY , STAT, free buffer requested , 392, 78.4,
4038, BILL_MY , STAT, CR blocks created , 10, 2,
4038, BILL_MY , STAT, physical reads cache prefetch , 5, 1,
4038, BILL_MY , STAT, shared hash latch upgrades - no wait , 375, 75,
4038, BILL_MY , STAT, calls to kcmgas , 376, 75.2,
4038, BILL_MY , STAT, redo entries , 9, 1.8,
4038, BILL_MY , STAT, redo size , 648, 129.6,
4038, BILL_MY , STAT, redo subscn max counts , 9, 1.8,
4038, BILL_MY , STAT, file io wait time , 4.3M, 860.97k,
4038, BILL_MY , STAT, data blocks consistent reads - undo records applied , 476, 95.2,
4038, BILL_MY , STAT, rollbacks only - consistent read gets , 1, .2,
4038, BILL_MY , STAT, cleanouts and rollbacks - consistent read gets , 9, 1.8,
4038, BILL_MY , STAT, immediate (CR) block cleanout applications , 9, 1.8,
4038, BILL_MY , STAT, commit txn count during cleanout , 9, 1.8,
4038, BILL_MY , STAT, cleanout - number of ktugct calls , 9, 1.8,
4038, BILL_MY , STAT, table scan rows gotten , 492, 98.4,
4038, BILL_MY , STAT, table scan blocks gotten , 10, 2,
4038, BILL_MY , STAT, heap block compress , 72, 14.4,
4038, BILL_MY , TIME, DB CPU , 60.99ms, 12.2ms, 1.2%, |@ |
4038, BILL_MY , TIME, sql execute elapsed time , 4.35s, 869.12ms, 86.9%, |@@@@@@@@@ |
4038, BILL_MY , TIME, DB time , 4.35s, 869.12ms, 86.9%, |@@@@@@@@@ |
4038, BILL_MY , WAIT, db file sequential read , 4.17s, 834.69ms, 83.5%, |@@@@@@@@@ |
4038, BILL_MY , WAIT, db file scattered read , 49.17ms, 9.83ms, 1.0%, |@ |
-- End of Stats snap 1, end=2012-05-21 22:22:16, seconds=5
...省略另两次的采样输出。
本文来自电脑杂谈,转载请注明本文网址:
http://www.pc-fly.com/a/jisuanjixue/article-30053-3.html
千玺是最棒的
一次次双降
badboy