How does SCSI RAID controller controls reads and writes on a RAID 0 disk array?
Assuming the stripe size of 64K and an insert is trying to write to a datafile that is striped on all three disks and a query is trying to read from tables and indexes that are striped on all three disks, how does the RAID controller handles the requests? Does it permit asynchronous read and write on all the disks in parallel? Where can I get info on this?
Hmm....... asynchronous commit feature? Good for performance.. but....
The obvious danger of asynchronous commits is that your DBMS is no longer ACID complient. Specifically, in using an asynchronous commit, you have traded durability for speed. An asynchronous commit returns before the data has actually been written to disk, so if the database crashes before some buffered redo is written to the logs, or a file I/O problem prevents the log writer from writing buffered redo to a non-multiplexed log, then the commit is lost.
Also see the last part of the metalink doc: 336219.1.
Last edited by Thomasps; 07-31-2006 at 11:57 AM.
Buy a useable storage array that meets your requirements for Availability, Performances and cost.
Just 4 disks is not enough to get even moderate availability/perfomance. See your own stats with High log sync wait event!
Click Here to Expand Forum to Full Width