OOSP: Opportunistic Optimization Scheme for Pod Deployment Enhanced with Multilayered Sensing
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe authors proposed an opportunistic pod placement optimization technique using multi-layer inspection to address the issues such as performance degradation and resource wastage in default scheduler, optimized for CPU and memory usage. The method analyzes multi-layer data to derive optimal placements based on pod coupling and dependencies, thereby improving resource utilization and performance. However, the manuscript has adopted a novel approach to prove its place, still some basic observations need proper justifications. I suggest the authors to work on the following issues:
1. The content of the abstract is structured well, and the importance of the research problem is clearly stated. However, the abstract has to be revised and concise. It is better to mention the all-validation metrics used to compare the performance of the proposed technique with that of state-of-the-art techniques
2. Introduction should be shortened by removing redundant descriptions and the major contributions of the proposed model should be discussed in detail (i.e.) describe the efficiency of opportunistic pod placement optimization technique in stepwise pattern with the corresponding data and the performance upgradation of a proposed model comparing with the existing model
3. In the related work section, the authors can include recent works on container scheduling and Kubernetes scheduling.
· Lai, W.K., Wang, Y.C. and Wei, S.C., 2023. Delay-aware container scheduling in kubernetes. IEEE Internet of Things Journal, 10(13), pp.11813-11824.
· Lavanya, J. and Kavi Priya, S., Multi-objective comprehensive container scheduling and resource allocation for container cloud using tuna swarm optimization algorithm. Journal of Intelligent & Fuzzy Systems, (Preprint), pp.1-17.
· Lai, W.K., Wang, Y.C. and Wei, S.C., 2023. Delay-aware container scheduling in kubernetes. IEEE Internet of Things Journal, 10(13), pp.11813-11824.
4. In Section 3.2, the authors haven’t discussed the Run Workload module of the overall system process flow depicted in Figure 2. A detailed description of the Run Workload module should be included, along with its corresponding identification of service behavior and pod behavior. This addition will provide a more comprehensive understanding of the system's operation and dynamics.
5. The authors have mentioned "YAML files" in the context of modifying the workload and executing the orchestration in Section 3.2. However, a detailed explanation of YAML files is necessary for a better understanding, especially for researchers in the cloud computing environment. Also, citing relevant manuscripts or references regarding YAML would provide further clarity and support.
6. The line 253 clarity is missing
7. The authors are requested to illustrate Algorithm 1 and Algorithm 2 in detail. Providing comprehensive explanations and stepwise descriptions of each algorithm will enhance the clarity for the researchers
8. The authors can include the mean threshold formula in threshold calculation for Pod Placement Optimization
9. Why the authors have considered only two performance metrics (average response time and request per second) to ensure the performance of the proposed model.
10. Is it possible to evaluate the resource utilization before and after deploying the optimization algorithm to ensure that the resources are efficiently used?
11. The authors can include latency and throughput to ensure the application-level performance of the proposed model
12. It's better to elaborate on the inferences of Figure 10 and Figure 11
13. The inferences of figure 12,13,14,15 to be enhanced by explaining the advantage of improved average response time and request per second
14. Improve the clarity of all figures
15. Check the typos and grammatical errors in the entire manuscript
16. The language of the paper should be revised, there are multiple errors and the text is often unclear
17. Conclusion section must be revised and precise
I would recommend a “Major revision” to enhance the quality of the article and would like to review an updated version of the manuscript based on the above-said comments.
Comments on the Quality of English LanguageCan be improved.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsRemarks:
1) A few concepts can be disputed and all the existing sentences are potentially too complicated so they should be made less complex. One might want to shift from complex to simple constructions, perhaps splitting long sentences into short ones.
2) It is beneficial to have figures and tables; however, it is also necessary to refer to them in the text more elaborately. To do so, make certain that every figure is given a discussion that elaborates on why the figure is important to the study context.
3) Some aspects missing from the manuscript are considering limitations on the experimental setting, e.g., the applicability of the presented approach in various cloud infrastructures.
4) Though the study reveals increments in special conditions, the work does not explore how the proposed technique can be implemented when the number of nodes and pod rises. Sometimes when the above concepts are scaled, several challenges may result in not being able to provide or sustain good throughput and turnaround.
5) Work bases the study on simulation approaches and utilizes real benchmark applications for the work. Unfortunately, it is still unclear how this stated method practicality for batches looks like in real-life cloud configurations dealing with rather complex workloads in the context of availability.
6) It is also important for the manuscript to contain a discussion on the scalability of the proposed algorithms so that it is possible to explain the level of feasible reachability of presented algorithms if the algorithms are to be implemented in a large scale.
7) Conclusions and Future work: The final section of the article, ‘Future work’, is rather suggestive but the authors could have included some research questions there or, at least, some application of the method which can focus, for example, real-time cloud systems.
8) Literature Review: The related work section is comprehensive, but seeing how the proposed method is ‘better’ or solves some of the existing issues in literature could have been done in greater detail. It may be useful to provide a table that would compare this work to other methods, therefore highlighting the contribution of the study.
9) The manuscript would also have been enhanced by a more detailed qualitative analysis of the algorithmic behavior (algorithmic complexity). In particular, explain what computations had to be performed to obtain the results of optimization algorithms and how these calculations change with the growth of the size of the system,
10) Though, the paper briefly mentions possible malicious activities, such a discussion of security implications could be more elaborate. What happens to pod placement security issues or how sensitive will a pod be to attacks that may be devised with regards to optimization? Kindly mention potential vulnerabilities or risk attacks on the optimization process.
"The proposed method reduced average response time by up to 11.5% and increased requests per second by up to 10.04% compared to the default Kubernetes scheduler." Line 9
"The dynamic nature of cloud-native systems requires continuous adjustments by system architects, increasing administrative fatigue." Line 50
"Additionally, administrators manually specifying resource requirements may result in inaccurate settings, negatively impacting the entire cluster and leading to various human errors." Line 171
"The Application Level in the Data Collection module collects workload data as shown in Figure 4." Line 214
"The saved pcap files are preprocessed to filter method calls invoked in the workload between nodes." Line 219
"In modern cloud environments, container orchestration tools are essential for managing workloads and services." Add "the".
"Additionally, administrators manually specifying resource requirements may result in inaccurate settings, negatively impacting the entire cluster and leading to various human errors."
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsThe authors have taken into consideration all remarks. Ultimately, the manuscript has been significantly improved.
However, the section of background & related work should be imroved. Include more Container Scheduling Techniques:
Some State-of-the-art methods are:
i) Prequal (Load Balancing for Distributed Systems).
ii) SLAM (Memory Optimization for Serverless Applications).
iii) Tuna Swarm Optimization (TSO).
(+) It would be usefull if you compare the proposed frame (OOSP) with other existing methods in terms of (e.g., memory, CPU usage).
-) Real-time Pod Scaling: Discuss how OOSP could scale in real-time. Explain this better.
Comments on the Quality of English LanguageMinor.
Author Response
Reply Letter
- Manuscript ID: sensors-3132164
- Manuscript Title: “OOSP: Opportunistic Optimization Scheme for Pod Deployment Enhanced with Multilayered Sensing”
- Authors: Joo-Young Roh, Sang-Hoon Choi, and Ki-Woong Park
September 21, 2024
Dear Reviewers of this manuscript:
We wish to submit our revised manuscript reflecting the reviewers’ comments regarding publication in SENSORS, titled “OOSP: Opportunistic Optimization Scheme for Pod Deployment Enhanced with Multilayered Sensing”
We would like to thank the reviewers for their time and valuable comments. We thoroughly revised the manuscript in accordance with your comments, and our responses are enclosed. All of your comments were helpful toward improving the quality of this paper.
Thank you.
Sincerely,
Prof. Dr. Ki-Woong Park (Corresponding Author)
Author Response File: Author Response.pdf
