Re: Netapps vs EMC -- db file scattered read

I wonder if the locality of data has anything to do with it.

If the read-ahead config of the EMC unit is optimized, you may be benefiting from cache hits on read ahead. Its possible that the stripe size on the EMC array is better matched to your OS and database configuration than the stripe size of the NetApp filer, making table scans more efficient. its possible that write anywhere, file later is making your scattered reads much more scattered as more latency could be introduced due to non-contiguous data.
do these units contain comperable disks, number of channels/paths? is the NetApp Filer connected point-to-point or via a longer path?

check Connor McDonald's site for scripts for determining the max setting for the db_file_multiblock_read_count. sounds like alot of "it depends".

-bdbafh Received on Tue Jan 25 2005 - 23:42:07 CST