平均每次最多是200到400个会话，但是总会在某些时候，会话逛涨，报如下错误：
我用的是ORACLE 11g
ORA-00020: 超出最大进程数 (2500)
ORA-20 errors will not be written to the alert log for
the next minute. Please look at trace files to see all
the ORA-20 errors.
Process PO10 submission failed with error = 20
Sun Jul 28 16:52:21 2013
Process PO11 submission failed with error = 20
Sun Jul 28 16:52:21 2013
Process PO12 submission failed with error = 20
Sun Jul 28 16:52:21 2013
Process PO13 submission failed with error = 20
Sun Jul 28 16:52:21 2013
Process PO14 submission failed with error = 20
Sun Jul 28 16:52:22 2013
Process PO15 submission failed with error = 20
Sun Jul 28 16:52:22 2013
Process PO16 submission failed with error = 20
Sun Jul 28 16:52:22 2013
Process PO17 submission failed with error = 20
Sun Jul 28 16:52:22 2013
Process PO18 submission failed with error = 20
Sun Jul 28 16:52:22 2013
Process PO19 submission failed with error = 20
Sun Jul 28 16:52:22 2013
Process PO20 submission failed with error = 20
Sun Jul 28 16:52:31 2013
Process J000 submission failed with error = 20
kkjcre1p: unable to spawn jobq slave process
Errors in file D:\APP\ADMINISTRATOR\diag\rdbms\orcl\orcl\trace\orcl_cjq0_11696.trc:
Process J000 submission failed with error = 20
kkjcre1p: unable to spawn jobq slave process
Errors in file D:\APP\ADMINISTRATOR\diag\rdbms\orcl\orcl\trace\orcl_cjq0_11696.trc:

在连接数突增的时候，查看连接进程状态，还有查看一下你在这个时间段有没有大的job。

也要看看有没有应用异常，有的时候因为应用异常，会阻塞其他进程，从而不能释放资源。
还有看一下在这个时间段有没有表的重定义操作。因为在11.2中有个bug（Bug 11834459: DBMS_REDEFINITION.FINISH_REDEF_TABLE MAY FLUSH THE SHARED POOL）。这个操作会导致会话被锁的问题。