The authors of this paper introduce a new minimal metric monitoring service (M3S) to evaluate cloud provider services, in support of optimal use of federated cloud resources.

All of the other components comprising the federated cloud management (FCM) infrastructure, including the generic meta-broker service, SALMon service monitoring, and cloud-broker, appear to be previously published works by these authors. There is some mention of extensions to this work, but since these are not described, I assume they are so minor as to not merit discussion. Therefore, the heart of what is new is M3S and what it measures, and the value of using its results.

M3S runs as a virtual machine (VM) within the specific cloud infrastructure, and measures VM deployment time, response availability (ping), and changes in response time when M3S is conducting fixed-size upload or download and running central processing unit (CPU) intense jobs.

The results demonstrated can only be understood in a testbed where the entire environment is carefully controlled. Given that the results shown are for use of M3S in federated and uncontrollable environments (real life), I fail to understand how any type of normal baseline behavior could be determined from which to measure subsequent variation.

In light of this, the actual conclusion reached by the authors--“The presented evaluation results show that both service reliability and responsiveness do vary over time and load conditions”--is underwhelming. Furthermore, the conclusion that “these measures can be used by our federated cloud management solution to select better execution environments for achieving a higher level of user satisfaction” is unsubstantiated.

This paper also suffers from many grammatical errors, a lack of references for previous work, and some unnecessary repetition.