表分析job比以前慢了4个小时

nan_dream

环境：
=========================
OS：HP-UX 11.23
DB：10.2.0.5.12【RAC】




问题：
=========================
3月25日晚上去掉飞康CDP设备（用于做磁盘镜像复制），发现表分析job完成需要比以前多4个多小时的时间【这个job以前也是每天都跑，表数据量没有很大的变化】

请问这个应该从哪里入手进行排查？【表分析语句请看附件】

谢谢！

Liu Maclean(刘相兵

最好有10046 trace，如果没有至少也要有对比当时的AWR和ASH

nan_dream

Liu Maclean(刘相兵 发表于 2014-3-29 15:39
最好有10046 trace，如果没有至少也要有对比当时的AWR和ASH

去掉飞康CDP之前I/O是25M/s，去掉之后是75M/s，按道理数据库应该是快了，为什么反而变慢了？？

gavintgh

ipad下载不了

sheep_yang678

FINDING 2: 28% impact (26685 seconds)
-------------------------------------
The throughput of the I/O subsystem was significantly lower than expected.

   RECOMMENDATION 1: Host Configuration, 28% benefit (26685 seconds)
      ACTION: Consider increasing the throughput of the I/O subsystem.
         Oracle's recommended solution is to stripe all data file using the
         SAME methodology. You might also need to increase the number of disks
         for better performance. Alternatively, consider using Oracle's
         Automatic Storage Management solution.
      RATIONALE: During the analysis period, the average data files' I/O
         throughput was 18 M per second for reads and 97 K per second for
         writes. The average response time for single block reads was 39
         milliseconds.

   RECOMMENDATION 2: Host Configuration, 0.85% benefit (826 seconds)
      ACTION: The performance of file /dev/vgoradata/rlv_yd_part_data27 was
         significantly worse than other files. If striping all files using the
         SAME methodology is not possible, consider striping this file over
         multiple disks.
         RELEVANT OBJECT: database file
         "/dev/vgoradata/rlv_yd_part_data27"
      RATIONALE: The average response time for single block reads for this
         file was 47 milliseconds.
你的75M/S怎么测试的？