Huh? Define Sand-bagging i/o?Quote:
Originally Posted by KenEwald
It all depends on how the system is layed out...
Sure I/O will be terrible if you share drives between the arch file systems and redo logs (raw, file system).... But If you don't and have dedicated devices for each, then what does it matter if occasionally I/O spikes for 10 seconds to generate an arch log?
It better to generate fewer archive logs to reduce checkpoints, but in the end this must be ballanced by many other factors(ie data guard, backups etc).
I guess all i'm saying is a 2M log file is generally a bad idea, no matter what the system. It might be ok for say a read only or a DW where all loads are direct path and very little redo is generated.
Hope this helps,