Here is another insight on Amazon's EC2 services that we found during our recent cloud monitoring experiments: For a couple of days we ran our
load test sensor for disk speed monitoring on instance stores (disk volumes which come with each instance) as well as on EBS volumes (Amazon's
Elastic Block Store) for all available instance types.
The results for a period of 48 hours are shown in this graph from PRTG Network Monitor:
The values are percentages of test time on instance store divided by test time on EBS store. E.g. for m1.small an average of 113% means that on average the disk test took 13% more time on the instance volume than on the EBS store (=the EBS store was faster).
Conclusion
Except for our m1.small instance (EBS was 13% faster) in our tests we could not find any significant difference between the two disk types.
Amazon say on their
website: "
The latency and throughput of Amazon EBS volumes is designed to be significantly better than the Amazon EC2 instance stores in nearly all cases."
We did not see a significant advantage of EBS volumes. Shall we blame the simplicity of our load test? Amazon continues: "
The exact performance will depend on the application [..], so the best measure is to benchmark your real applications against the volume.".
We must agree for now - our simple load test sensor did not show any differences between both options. We should revisit this topic soon.
Disclaimer
As I said, our test program is only a very simple test. It is no substitute for application-specific "real world" tests that you should run on the instance types if you consider moving applications onto EC2 or any other cloud hosting service.
Test Notes
Our tests were run on April 5th and 6th 2009 on Windows based instances in the EC2 region "US East Coast" and availability zone "1c", each instance had its own EBS volume attached with a size of 1 GB.