Poor IO performance

Hi,
we recently started to migrate our Mongodb cluster but we are experiencing a really poor IO performance.
We run some tests to benchmark the iops and the results are even worse than those run in a standard Amazon EBS volume. In fact, the iops count is far below from the advertised 500 iops. We also tried to set a raid0 disk array to see if we could better results but didn't see any improvement at all:

sudo fio --name=test --rw=read --direct=1 --size=128M

Azure VM:
READ: io=131072KB, aggrb=721KB/s, minb=721KB/s, maxb=721KB/s, mint=181582msec, maxt=181582msec

AWS:
READ: io=131072KB, aggrb=7730KB/s, minb=7915KB/s, maxb=7915KB/s, mint=16956msec, maxt=16956msec

And if we try a read&write test, the result is even worse:

Azure VM:
READ: io=65572KB, aggrb=299KB/s, minb=299KB/s, maxb=299KB/s, mint=218668msec, maxt=218668msec
WRITE: io=65500KB, aggrb=299KB/s, minb=299KB/s, maxb=299KB/s, mint=218668msec, maxt=218668msec

Obviously with these numbers our service is completely unusable...

Is there any thing we are missing?
The OS is Ubuntu 13.04, the VM type is A6(4 cores, 28GB ram) and the tested device is a disk array of 6 disks in raid0 (we tried this to see if we could get a performance boost as recommended in other articles)

Thank you in advance