The Main Point
The point that the authors of the article have made is that the IT industry requires energy-efficient solutions. They chose a particular sphere of cloud computing as the primary focus. They report that data centers used in cloud-based solutions can be made more energy-efficient by migrating virtual machines from server to server (Gul, Khan, Mustafa, Khalid, & Khan, 2019). They proposed several different solutions for migration with various levels of energy efficiency and quality of service.
Many VM migrations must be performed to keep the data center as energy efficient as possible, and some resources must be allocated to handle unexpected workload spikes. The article discusses six solutions, four of which have been put forward by the authors and two taken from existing energy-saving schemes. CRT and ACRT techniques are based on CPU and RAM availability and migrate VMs among servers based on their load.
SCRT and SACRT are the same technique with Service Level Agreement-aware dynamic upper threshold of available resources applied to them. ECREW and SCREW are the existing solutions, with the latter having the same SLA-awareness distinction. The most energy-efficient solution, ACRT, produced the most SLA violations. SCREW, the existing solution that produced the least SLA violations, is also the most energy-consuming. The results show that energy efficiency and quality of service have an inverse relationship.
Things I Learned
The article explains how the CPU and RAM of the servers in data computing solutions have different but essential roles. Both of these components must be considered in creating solutions, as both have different performance measures and energy needs. Also, I have learned that virtual machines are not static and migrate around servers to save energy. That may explain some performance irregularities we sometimes experience while using cloud solutions.
Questions to the Authors
The authors have shown that energy efficiency and quality of service are, to some degree, mutually exclusive. The solutions presented by them are environmentally-driven and are aimed at reducing carbon dioxide emissions. Is the considerable reduction in service quality justified to reduce emissions by a mere couple percent? Are there other green solutions, perhaps, in more abundant clean energy, that could serve as an alternative to considerably handicapping a vital knowledge-generating resource? Is it fair and ethical to impose these consumer-unfriendly solutions on de facto monopolies such as Google?
Reference
Gul, B., Khan, I. A., Mustafa, S., Khalid, O., & Khan, A. R. (2019). CPU–RAM-based energy-efficient resource allocation in clouds. The Journal of Supercomputing, 75(11), 7606–7624.