A Model-Driven Architecture for Automated Deployment of Microservices
Round 1
Reviewer 1 Report
The authors have enhanced their previous work and proposed a model-driven approach for the automated deployment of microservices to minimize the execution cost and communication costs among the microservices. The paper is well-organized and technically solid.
The comments are as below.
- The related work can be enhanced, for example, the following paper should be added so as to make a comprehensive literature review
- It is no need to put the code (Fig 2) into the context. Describe it in details instead.
- The complexity of the proposed algorithm should be elaborated.
Author Response
1. The related work can be enhanced, for example, the following paper should be added so as to make a comprehensive literature review
Answer: We have enhanced and reorganized the related work by adding recommended articles and other related articles.
2. It is no need to put the code (Fig 2) into the context. Describe it in details instead.
Answer: Fig.2 is removed from the article. Besides, the concept of Emfatic declarations and corresponding metamodel is detailed.
3. The complexity of the proposed algorithm should be elaborated.
Answer: The complexity of the proposed algorithm is elaborated in the Evaluation section.
Reviewer 2 Report
In this paper, the authors proposed a model-driven approach for the automated deployment of microservices. Through a case study, it was shown that the proposed method can be effectively applied. After reading the manuscript, I have the following questions and concerns. Please check the following concerns.
- On page 2, line 74, there is MDE, isn't it MDD?
- The figure in Appendix A is difficult to read due to poor legibility. If there is a way to improve it, it needs to be corrected.
- There are parts of the bibliography that are incorrect. For example, check line number 121.
- The total amount of memory required in Table 1 is incorrect.
- The online book shopping shown in the case study seems to be relatively small. Is the level sufficient to verify the proposed technique? Please clarify this.
- Related to Table 2, doesn't CCDetails have a Publish/Subscribe relationship to other services such as User, Customer, Seller, and so on?
- If possible, it will be necessary to bring each figure in the Appendix to the main text and explain it in more detail.
- There are too many typos in the manuscripts. Please proofread it again and fix it. For example %14.3 -> 14.3% on line number 281, and %23 -> 23% on line number 282.
Author Response
In this paper, the authors proposed a model-driven approach for the automated deployment of microservices. Through a case study, it was shown that the proposed method can be effectively applied. After reading the manuscript, I have the following questions and concerns. Please check the following concerns.
1. On page 2, line 74, there is MDE, isn't it MDD?
Answer: Thank you for this careful inspection. We corrected MDE as MDD.
2. The figure in Appendix A is difficult to read due to poor legibility. If there is a way to improve it, it needs to be corrected.
Answer: Appendix A is scaled and the images become clearer.
3. There are parts of the bibliography that are incorrect. For example, check line number 121.
Answer: Bibliography part is reorganized and all spell checks are corrected using Endnote Journal of Applied Science style.
4. The total amount of memory required in Table 1 is incorrect.
Answer: The second column of Table 1 represents the required memory per microservice instance. So total amount of memory is calculated by multiplying the number of instances* required memory. So the total amount of memory is 22.800 MB for the case study. To clarify the calculation, the column named is corrected as “Required memory per instance”.
5. The online book shopping shown in the case study seems to be relatively small. Is the level sufficient to verify the proposed technique? Please clarify this.
Answer: In this study, we aimed to explain the modeling part of the approach. We focused on the logical infrastructure of an automated microservice deployment approach by using model-driven architecture at the design phase. The plug-ins used in the model transformation process required for the metamodels to realize the approach and the implementation of the proposed model-driven architecture on the Eclipse Modeling Tool are detailed. In our previous study [1], we presented a systematic approach for enabling the deployment of microservices to limited capacity resources, and we detailed deployment process steps. Besides, we validated the approach by using a case study named Taxi-Hailing System (Uber) which consists of thousands of microservices. So we preferred a small-scaled online book shopping to evaluate the architecture for this study. The proposed model-driven architecture can be used effectively for both small-scale and large-scale microservice-based applications.
6. Related to Table 2, doesn't CCDetails have a Publish/Subscribe relationship to other services such as User, Customer, Seller, and so on?
Answer: Publish/subscribe relationships can be reconstructed according to the designer. To evaluate the proposed model-driven architecture, the authors created a base publish/subscribe relations, and CCDetails (debit card) data is not related to any service for this case study. In this scenario, Order service uses CardDetails since users pay with the credit card as a payment method.
7. If possible, it will be necessary to bring each figure in the Appendix to the main text and explain it in more detail.
Answer: All metamodels mentioned in Appendix A are separately explained in our previous study [1]. The relationships between metamodels are illustrated in Appendix A.
8. There are too many typos in the manuscripts. Please proofread it again and fix it. For example %14.3 -> 14.3% on line number 281, and %23 -> 23% on line number 282.
Answer: Thank you for this careful inspection. All typos have been corrected in the manuscript.
Round 2
Reviewer 1 Report
The authors have address the comments properly.
Reviewer 2 Report
All comments are resolved. I do not have further concerns or questions.
