【云迁移论文笔记】Cloud Migration Research:A Systematic Review

Cloud Migration Research:A Systematic Review

Author Info:

Pooyan Jamshidi

PhD

Postdoctoral Researcher

Dublin City University · School of Computing

Major: model-driven software architecture evolution

PS:

This paper is the first SLR(Systematic Literature Review) on Cloud Migration

INDEX

1 Introduction

(Main: introduce state-of-the-research)

2 Background And Related Work

(Main: background and related research to position contributions of this work)

2.1 SOA versus Cloud Migration: Similarities and Differences

2.2 Software Migration and Modernization

2.3 Secondary Studies on Legacy to SOA Migration

2.4 The Needs for a Secondary Study on Cloud Migration

3 Research Methodology

(Main: research methodology,research questions,and scope)

3.1 Planning the Review

3.2 Conducting the Review

4 A Framework For Classification And Comparison Of Cloud Migration Research

(Main: a reference model for state-of-the-research and a characterization scheme for cloud migration)

4.1 Cloud-RMM: A Cloud Migration Reference Model

4.1.1 Existing SOA Migration Frameworks

4.1.2 Cloud-RMM Reference Model

4.2 A Characterization Framework for Cloud Migration

5 Results

(Main: the results of the systematic review)

5.1 Overview of the Primary Studies

5.1.1 Temporal Overview of Studies

5.1.2 Publication Fora

5.1.3 Active Research Communities

5.1.4 Research Methods for Cloud Migration

5.2 A Taxonomical Classification and Comparison of Cloud Migration Methods and Techniques

5.3 Primary Drivers for Cloud Migration

5.4 Migration Types

5.4.1 Methods, Techniques and Tool Support for Cloud Migration

6 Research Implications And Future Dimensions

(Main: the main findings,implications,and trends)

6.1 Maturity of Cloud Migration Research

6.2 Research Trend and Future Directions

6.3 Benefits for Researchers and Practitioners

7 Threats To Validity

(Main: an analysis of its limitations)

8 Conclusions

(Main: concludes the paper)

1. Introduction

this SLR with the primary objective to identify,taxonomically classify,and systematically compare the existing research focused on planning,executing,and validating migration of legacy systems toward cloud-based software.More specifically,aim to answers the following questions by conducting a methodological review of existing research:

u What are the main practical motivations behind legacy migrations to the cloud?

u What are the existing tasks,methods,and techniques to enable migration of legacy on-premise software to the cloud?In addition,what is the available tool support?

u What are the existing research themes? And what should form future research dimensions in legacy-to-cloud migration?

2. Background And Related Work

2.1 SOA(Service-Oriented Architecture) VS Cloud Migration: Similarities and Differences

2.2 Software Migration and Modernization

Simple introduced Software Migration: Software migration is a process of moving or adapting an existing system from one operating environment to another.So-called “Modernization” means functional equivalence has been reached between the initial and the transformed system,the migrated systems can be aligned more closely to current and future business needs through adding new function-ality.

2.3 Secondary Studies on Legacy to SOA Migration

2.4 The Needs for a Secondary Study on Cloud Migration

None of the retrieved secondary studies was related to any of our research questions.Considering the importance of cloud migration and the relative maturity of this field (see Section 6.1),a con-solidation of existing evidence on legacy-to-cloud migra-tion is timely.

3. Research Methodology

Summarize: planning,conducting and documenting.

3.1 Planning the Review

Step 1—Identify the needs for SLR

Step 2—Specifying the research questions

Step 3—Define and evaluate review protocol

3.2 Conducting the Review

Step 1—Select primary studies

Initial selection

Final selection

Qualitative assessment of included studies

Based qualitative assessment on factors like general assessment (G) and specific assessment (S)

Steps 2 and 3—Data extraction and synthesis

4. A Framework For Classification And Comparison Of Cloud Migration Research

4.1 Cloud-RMM: A Cloud Migration Reference Model

The aim of this reference model is to identify areas that existing work is concentrated on and areas not covered.By identifying these areas,we can identify trends and directions of research on cloud migration.

4.1.1 Existing SOA Migration Frameworks(SOA-MF)

The framework involves three processes:

1) reverse engineering,

2) transformation,

3) forward engineering.

SOA Evaluation framework: Khadka developed a migration framework for legacy-to-SOA evolution.However,they do not consider migration validation,an important step after performing and implementing migration.In addition,they do not consider crosscutting activities such as governance.

SMART: The SEI has developed a SOA migration framework that helps organizations to analyze legacy systems and migrate them to a service-based system.This framework has shortcomings.They only concentrate on early stages and do not cover forward engineering activities.

4.1.2 Cloud-RMM(Cloud-Reference Migration Model) Reference Model

Based on previous experience,we adopt situational method engineering to consolidate existing frameworks in cloud migration.

Process I.Migration planning.

Process II.Migration execution

Process III.Migration evaluation

Process IV.Crosscutting concerns

Implications of Cloud-RMM.This reference model helps to conceptualize a process-centric view for state-of-the-research and literature classification into processes and associated tasks.Cloud-RMM can distinguish solutions for migration.

4.2 A Characterization Framework for Cloud Migration

TABLE 7

5. Results

5.1 Overview of the Primary Studies

5.1.1 Temporal Overview of Studies

5.1.2 Publication Fora

TABLE 14

5.1.3 Active Research Communities

5.1.4 Research Methods for Cloud Migration

Contribution type.FIG.6 Solution Proposal(65%),Experience reports(22%),Evaluation research(13%).

Evaluation method.Fig.7 suggests that controlled experiments and case studies are the most common evaluation method adopted.

5.2 A Taxonomical Classification and Comparison of Cloud Migration Methods and Techniques

5.3 Primary Drivers for Cloud Migration

ü Operational cost saving (48 percent of the studies).

ü Application scalability (26 percent of the studies).

ü Efficient utilization of resources (21 percent of the studies).

5.4 Migration Types

Ø Type I (replace): Data and/or business tiers have to be migrated to the cloud stack.We could not identify any evidence of this type of migration in the selected studies

Ø Type II (partially migrate):migrates some of the software system’s components to the cloud(34 percent of the studies have metioned)

Ø Type III (migrate the whole application stack): the easiest way of migration,where the whole application is monolithically encapsulated in one or more virtual machines running on the cloud.This type of migration needs no adaptation assuming the application stack can be ported “as is” into a virtual machine.(21 percent of the studies have metioned)

Ø Type IV (cloudify): the most complete migration (43 percent of the studies have metioned)

5.4.1 Methods,Techniques and Tool Support for Cloud Migration

Migration decision support.

Migration techniques.

Experience and lessons learned.

6. Research Implications And Future Dimensions

6.1 Maturity of Cloud Migration Research

6.2 Research Trend and Future Directions

u The growing maturity level.

u The needs for a comprehensive migration framework.

u Solutions to address crosscutting concerns.

u Automated support in migration process.

u The needs for architectural adaptation support.

u The needs for self-adaptive cloud-enabled systems.

6.3 Benefits for Researchers and Practitioners

7. Threats To Validity

8. Conclusions

Ø The field is still in a formative stage(in early states of maturity)

Ø a need for a concrete migration framework to enable systematic migration to the cloud.

Ø a lack of tool support for enhancing cloud migration.

Ø a lack of evidence for architectural adaptation support and self-adaptive cloud systems