Re: doubt about ASM

I think we need to investigate more on this. Yes, Oracle docs are not saying more on this even
I tried to search Oracle Metalinks but could not get this info.

I did some more test and wanted to share with you.

I was having 3 Disks ---
I created two diskgroup. One is external redund. and second one using normal redund using two failgroup.
I created two tablespace and placed them into these two diff. diskgroups.
Then I created two table and placed with these diff. tablespaces.

So, Finally layout was like this.

 Table Tablesapce Diskgroup | Failgroup Disks

Table_1 -> tbs_ext -> DG_external --->> Disk1

Table_2 -> tbs_nor -> DG_normal / - Failgroup_1 -> Disk2

                                   \  - Failgroup_2  -> Disk3

I was having expression like if I removed Disk1 and Disk3 from os level then I will get
error for Table_1 as it is in Dg-external and no mirroring is being used and select for table_2
will get the data.

But after removing both disks [ Disk1 and Disk3 ] still I was able to select data from both tables
I flushed the buffer cache but still was able to select the data.

Do not know where I was wrong ** ?? ** .

One more case :

I use pl/sql and setup a loop of 10000 and insert and commit in table_1

and not the timing and I did same for table_2 [ which is in Normal_redundancy ]
 But both the time Insert was done almost same time. I noticed time in milliseconds.

I was hoping inserts for normal disk would be slow as oracle has to write in both failgroups.

I know mirroring is being done on extent level and if DBWR is not responsible for this and anyother process is doing this, But still it should show slow performance for normal Diskgorup.

Now, while typing one thing came in my mind that my commit  was invoking LGWR not DBWR.I think now I should change my core  and add one more line "alter system flush buffer cache after 100 records" . Then see.. what happend ..

Anyway -- Thank you very much for your help.

Regards,
Jagjeet Singh Received on Tue May 09 2006 - 08:09:29 CDT