I recently discovered that if I disable read ahead via writing a 0 to /sys/class/block/XXX/queue/read_ahead_kb that regardless of issuing a large read request (512kb for example), the actual i/o on the wire is fragmented into multiples of 4kb (or whatever the read ahead value is set to if non 0) causing far lower than expected throughput. I assumed this must be a bug, so I tested 2.6.37 as well as 3.4, and confirmed identical behavior on each. Is this the expected result? -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
