Why the I/O rate difference between an LV and PV?

Question

I have been doing some benchmarks reading with dd from both the LVM device and the PV device and I got some unexpected results:

$ sync ; echo 3 > /proc/sys/vm/drop_caches; dd bs=1M count=1024 if=/dev/sdc of=/dev/null
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 2.06466 s, 520 MB/s
[~]
$ sync ; echo 3 > /proc/sys/vm/drop_caches; dd bs=1M count=1024 if=/dev/mapper/vg-data of=/dev/null
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 9.07621 s, 118 MB/s
[~]

The difference is consistent on multiple runs, why is there such a difference and what of the transfer rate should be expected from applications reading from the corresponding FS?

Answer 1

From an IBM guide on LVM: "The mechanisms for defining and expanding logical volumes attempt to make the best possible default choices. However, satisfactory disk-I/O performance is much more likely if the installer of the system tailors the size and placement of the logical volumes to the expected data storage and workload requirements. "

see http://pic.dhe.ibm.com/infocenter/aix/v7r1/index.jsp?topic=%2Fcom.ibm.aix.prftungd%2Fdoc%2Fprftungd%2Fdisk_preinstall_guidelines.htm and http://pic.dhe.ibm.com/infocenter/aix/v7r1/index.jsp?topic=%2Fcom.ibm.aix.prftungd%2Fdoc%2Fprftungd%2Flog_vol_disk_io_perf.htm