Which RDS instance type has better IO: Extra Large DB Instance or High-Memory Extra Large Instance

I am getting IO performance concerns with Amazon RDS using mysql 5.5.12. There are 2 instance types that are similar and close in price:

Extra Large DB Instance: 15 GB of memory, 8 ECUs (4 virtual cores with 2 ECUs each), 64-bit platform, High I/O Capacity (MySQL DB Engine Only) ($0.88 per hour)

High-Memory Extra Large Instance 17.1 GB memory, 6开发者_StackOverflow中文版.5 ECU (2 virtual cores with 3.25 ECUs each), 64-bit platform, High I/O Capacity ($0.65 per hour)

Does anyone know if the Extra Large DB has faster IO than High-Memory Extra Large Instance?

m1.xlarge instances (the 64-bit/15GB you mention) get a NIC all to themselves and therefore have the best network (and by extension, EBS) performance. With striping you can sustain 1500 disk iops, as oppossed to the standard 100 iops.

When used in RDS, you will see highest iops by setting your storage size to at least 300GB, which triggers back-end striping.

If you need more RAM or CPU, only the c1.xlarge and m2.4xlarge instance types have the full NIC to themselves- they would also required 300GB storage allocation to take advantage of it.