Re: Row Migration/Rowchaining

  • From: Dan Norris <dannorris@xxxxxxxxxxxxx>
  • To: maheswara.rao@xxxxxxx
  • Date: Fri, 11 Apr 2008 14:56:26 -0500

My interpretation of this information is one or more of the following:
  • You have other chained rows in tables that are not analyzed (and therefore wouldn't be listed in your dba_tables query). If this is the case, then v$sysstat is including reads of chained rows in tables that are not yet analyzed.
  • If you're estimating stats, dba_tables won't have accurate chain_cnt--definitely not for migrated rows. Check sample_size in dba_tables to be sure.
  • There were other chained rows that are now gone.
  • Or, you really have touched those 101 rows 2.3 million times since instance startup.

maheswara.rao@xxxxxxx wrote:
Thank you Dan and Joe.

I checked dba_tables.  The query result shows 101 rows.

>From v$sysstat, I am getting 2,327,441.

Does v$sysstat contains other info when query for 'table fetch continued row'?

Please see below queries.

  1  SELECT owner, table_name, chain_cnt
  2    FROM dba_tables
  3*   WHERE chain_cnt > 0
SQL> /

OWNER                          TABLE_NAME                      CHAIN_CNT
------------------------------ ------------------------------ ----------
REPORTER                       REP_AUDIT_SEVERITY                     18
REPORTER                       REP_AUDIT_ACK                          83

= = = =

SQL> SELECT name, value
       FROM v$sysstat
       WHERE name = 'table fetch continued row';

NAME                               VALUE
---------------------------------- --------
table fetch continued row          2327441

Other related posts: