Generic Refactoring Methodology for Cloud Migration

Position Paper

Manoj Kesavulu, Marija Bezbradica and Markus Helfert

Lero – Irish Software Research Organization,

School of Computing, Dublin City University, Dublin, Ireland

Keywords: Generic Architectural Refactoring, Cloud Migration, Service-oriented Architecture, Cloud Platform.

Abstract: Cloud migration has attracted a lot of attention in both industry and academia due to the on-demand, high

availability, dynamic scalable nature. Organizations choose to move their on-premise applications to adapt to

the virtualized environment of the cloud where the services are accessed remotely over the internet. These

applications need to be re-engineered to completely exploit the cloud infrastructure such as performance and

scalability improvements over the on-premise infrastructure. This paper proposes a re-engineering approach

called architectural refactoring for restructuring on-premise application components to adopt to the cloud

environment with the aim of achieving significant increase in non-functional quality attributes such as

performance, scalability and maintainability of the cloud architectures. This paper proposes, when needed to

migrate to cloud, the application is divided into smaller components, converted into services and deployed to

cloud. The paper discusses existing issues faced by software developers and engineers during cloud migration,

introduces architectural refactoring as a solution and explains the generic refactoring process at an

architectural level.

1 INTRODUCTION

Cloud computing has become a buzzword in the IT

industry today and it is regarded as the most

influential technology in the present time. Given the

long term benefits of adopting to this promising

technology, many industries decided to migrate their

on-premise applications to cloud. Cloud migration is

defined a process of partially or completely moving

an organization’s digital assets, services, IT resources

or applications to a cloud platform with an intent to

significantly improve various quality attributes such

as performance, scalability, maintainability and

similar (Pahl et al. 2013). Moving an application

using “lift-and-shift” approach, (which is moving it

as it is), moves all the existing issues associated with

it to cloud and it is likely to introduce new concerns

and challenges. An alternative approach to move

legacy on-premise application to cloud is “Big Bang”

rewrite, (where the application is built in cloud from

scratch), is not only a time-consuming process but is

also extremely risky and will likely end in failure.

Another recent option is to follow service-oriented

approach where the target application is divided into

smaller components and converted into services

comprising of core business functions, before its

migration to the cloud. this also involves making sure

that new features are exploited and the entire

application will continue to working in the new

environment (Garg et al. 2016).

Migrating an application to a cloud platform is

difficult, costly and error-prone (Kwon & Tilevich

2014). A major part of the IT system are applications

which are integrated and support core business

process and services, many of which are used for

utility needs, and these are non-core applications,

meaning that components of these applications are

loosely coupled. These application components need

to be re-engineered before deploying to cloud for

better use of the services offered by the cloud

platform. There are scenarios where an application is

deployed over more than one cloud platform which is

termed as multi-cloud deployment (Jamshidi et al.

2015).

This paper presents a generic architectural

refactoring approach that facilitates the process of

transforming on-premise applications to use cloud-

based services. The paper is structured as follows:

Section 2 gives an overview of the cloud platform.

Section 3 discuss cloud migration and identifies

664

Kesavulu, M., Bezbradica, M. and Helfert, M.

Generic Refactoring Methodology for Cloud Migration - Position Paper.

DOI: 10.5220/0006373106920695

In Proceedings of the 7th International Conference on Cloud Computing and Services Science (CLOSER 2017), pages 664-667

ISBN: 978-989-758-243-1

issues at an architectural level. Section 4 introduces a

generic refactoring methodology and at last Section 5

describes conclusions and future work.

2 CLOUD PLATFORM

Traditional IT (Information Technology) aligns

resources according to the way the applications are

deployed within dedicated infrastructure and data

storage to fulfil business requirements. Cloud

computing (CC) has emerged as a computing

paradigm with benefits such as high scalability,

reduced IT costs, on demand self-service, pay-as-

you-go price models, elasticity in provision

computing resources. It achieves this by varying

workload so organisations became attracted to move

their on-premise systems to the cloud.

CC can be mainly divided into three service layers

as shown in Figure 1 below:

Figure 1: Cloud Service Layers (Petrolo et al. 2014).

 Infrastructure as a Service (IaaS) offers

computing resources, both physical and virtual,

for processing and storage.

 Platform as a Service (PaaS) offers development

environment for software developers to write

their applications on a particular platform

without worrying about the underlying hardware

infrastructure.

 Software as a Service (SaaS) offers software

applications that can be accessed and used by the

end-users.

In addition to the layers, some other layers are

introduced namely Data as a Service, Everything as a

Service, Network as a Service, Things as a Service

and Sensing and Actuation as a Service (SAaaS) and

so on. CC has five deployment models (1) Private

cloud (2) Public Cloud (3) Hybrid Cloud (4)

Community Cloud (Marinos & Briscoe 2009) and (5)

Multi-Cloud (Paraiso et al. 2012). With the the

abundance of options to be considered for the

selection of a suitable cloud service and deployment

model, organization has to decide on the following

aspects (Garg et al. 2016):

 Selection of Deployment Model

 Selection of Service model

 Selection of Appropriate Service Package

Upon making a suitable decision, the target

application should be re-engineered to fit the selected

cloud platform adhering to the chosen deployment

and service model. An important pre-requisite to be

fulfilled, while transforming application components

serving the underlying business functions, is that they

preserve the external behaviour after the migration

process. According to a survey conducted by Kratzke

& Quint (2017), cloud specific design methodologies

are yet to be designed and developed. This paper

introduces an early stage generic refactoring

methodology which guarantees successful

application of architectural refactoring while

migrating on-premise applications to cloud. In the

next section, we discuss cloud migration process in

more details.

3 CLOUD MIGRATION

Cloud migration is the process of moving

applications, services, code, and business logic

deployed on the on-premise infrastructure to cloud

platforms with the aim of achieving significant

improvement in performance, scalability and cost

reduction. There are various methodologies and

approaches discussed in literature regarding cloud

migration. Rowe, Brinkley and Tabrizi (2013)

designed a generic methodology for the migration of

legacy system to Cloud Platform describing the

following steps: (1) Architectural representation of

the legacy application; (2) Redesign the architecture

model; (3) MDA transformation; (4) Web service

generation; (5) Web service-based invocation of

legacy functionalities; (6) Selection of suitable Cloud

Computing Platform; (7) Web service deployment in

the service cloud. Considering the steps (2) and (3),

where the original architecture model is redesigned to

identify services that can be provided in a SaaS

architecture as shown in Figure 2 below:

Figure 2: Refactoring with Architecture Driven

Modernization (Rowe et al. 2013).

Generic Refactoring Methodology for Cloud Migration - Position Paper

665

This methodology explains the prerequisite steps

followed in the refactoring process rather than the

actual refactoring process.

The authors Kwon and Tilevich (2014) describe

an automated transitioning approach to cloud-based

services and discuss three scenarios: moving a part of

application functionality to cloud, adding new (fault-

tolerance) functionality and switching an application

to use an alternate cloud-based service. These three

scenarios use cloud refactoring approaches as a

solution. However, in a real-world situation these

approaches can only be applied to the scenarios

discussed or similar scenarios and do not serve as a

generic refactoring approach which could be applied

to all cloud migration scenarios. Hence there is a

crucial need to derive a generic refactoring process.

4 GENERIC REFACTORING

METHODOLOGY

Refactoring is a process of changing internal design

of the system while preserving its external behaviour

(Fowler 2002). The goal of refactoring is to improve

a certain quality while preserving others. It is a

bottom-up process which helps clean-up inconsistent

or insufficient design decisions which can be applied

for design artifacts, models, documents, UML

diagrams, processes and architectures (Stal 2007).

There have been numerous examples where

refactoring is applied in different cases and scenarios

with migrating legacy applications to cloud platforms

(Zimmermann 2016; Kwon & Tilevich 2014; Rowe

et al. 2013; Chauhan & Babar 2012; Schmidt et al.

2012). Refactoring can be simple – move, add,

rename, remove, pullup, substitute and so on and

complex – combination of two or more simple

refactorings.

The process of refactoring can be described as

follows (Figure 3):

Figure 3: Generic refactoring steps (Kesavulu, et al., 2016).

 Identify architectural bad smell: an architectural

bad smell is commonly (although not always

intentionally) used set of architectural design

decisions that negatively impacts system quality

(Stal 2007). Another definition for architectural

bad smell is the observation or the suspect that

something in architecture design and its

implementation is no longer adequate (i.e., good

enough) under the actual requirements and

current constraints for the system (Zimmermann

2016). Some of the examples are: high coupling

between subsystems or a data store which cannot

handle concurrent queries in reasonable

time(Zimmermann 2016); the former example is

a general architectural bad smell whereas the

latter is a cloud specific bad smell. Though there

are cloud specific bad smells and corresponding

architectural refactoring is available in the form

of catalogues, cloud specific techniques for

identification of smells and tools for application

of refactoring are deficient. Our future work

includes exploration of cloud specific bad smell

detection techniques.

 Choose Architectural Refactoring(s): Choosing

an appropriate architectural refactoring or a set of

refactorings to apply. Almost every bad smell is

associated with a potential architectural

refactoring. But there are various aspects to be

considered while choosing a refactoring such as

each refactoring should be applied incrementally

in small steps and specificity to cloud

environments.

 Check for potential collateral damage: A

refactoring may trigger other design

transformations due to existing dependencies

between application (or service) components,

some of which may be dangerous.

 Define Invariants, Pre and Post Conditions:

Before applying any refactoring, define

necessary steps to be taken to verify the

preservation of behavior of the system after

applying the refactoring.

 Apply Refactoring: The suitable target

refactoring chosen in step 2 is applied to the

legacy system (application) architecture

incrementally. Some of the refactorings are

Virtualize Server for IaaS layer, Swap

Messaging Provider for PaaS layer of the cloud

and so on.

 Check Invariants, Pre and Post Conditions: The

Invariants, Pre and Post conditions defined in

step 3 are verified which proves that the behavior

of the target system is preserved and the

refactoring is successfully applied.

This process can be considered as a generic

architectural refactoring methodology which can be

CLOSER 2017 - 7th International Conference on Cloud Computing and Services Science

666

used to verify the successful application of

refactoring to any cloud migration scenario.

5 CONCLUSIONS AND FUTURE

WORK

This paper introduces the cloud platform, discuss its

advantages over traditional IT platform and the need

for cloud migration and issues identified. The paper

proposes a generic architectural refactoring

methodology and introduces its steps while

explaining in detail. This methodology can be used to

verify successful application of refactoring process in

various scenarios of cloud migration.

The future work to this project would be to

develop a concept of detailed refactoring techniques

which includes methods to identify architecture

smells including exploration of existing architectural

smells identification and behaviour preservation

techniques; tools to recommend and apply

architectural refactoring specific to cloud service

architectures using a case study or a real-world

example. Evaluation of the methodology will be

conducted after choosing a suitable cloud

environment.

ACKNOWLEDGEMENTS

This work was supported with the financial support

of the Science Foundation Ireland grant 13/RC/2094

and co-funded under the European Regional

Development Fund through the Southern & Eastern

Regional Operational Programme to Lero - the Irish

Software Research Centre (www.lero.ie).

REFERENCES

Chauhan, M.A. & Babar, M.A., 2012. Towards Process

Support for Migrating Applications to Cloud

Computing. 2012 International Conference on Cloud

Computing and Service Computing (Csc), pp.80–87.

Fowler, M., 2002. Refactoring: Improving the Design of

Existing Code. In D. Wells & L. Williams, eds. Extreme

Programming and Agile Methods --- XP/Agile Universe

2002: Second XP Universe and First Agile Universe

Conference Chicago, IL, USA, August 4--7, 2002

Proceedings. Berlin, Heidelberg: Springer Berlin

Heidelberg, p. 256. Available at:

http://dx.doi.org/10.1007/3-540-45672-4_31.

Garg, R., Heimgartner, M. & Stiller, B., 2016. Decision

support system for adoption of cloud-based services.

CLOSER 2016 - Proceedings of the 6th International

Conference on Cloud Computing and Services Science,

1(Closer), pp.71–82. Available at:

https://www.scopus.com/inward/record.uri?eid=2-

s2.0-

84979743405&partnerID=40&md5=9ae740659dbd92

71f229e7cc1feaaf05.

Jamshidi, P. et al., 2015. Cloud Migration Patterns: A

Multi-cloud Service Architecture Perspective. In pp. 6–

19. Available at: http://link.springer.com/10.1007/978-

3-319-22885-3.

Kesavulu, M., Helfert, M. & Bezbradica, M., 2016.

Towards Refactoring in Cloud-Centric Internet of

Things for Smart Cities. Dublin, s.n.

Kratzke, N. & Quint, P.-C., 2017. Understanding Cloud-

native Applications after 10 Years of Cloud Computing

- A Systematic Mapping Study. Journal of Systems and

Software, (January).

Kwon, Y.W. & Tilevich, E., 2014. Cloud refactoring:

Automated transitioning to cloud-based services.

Automated Software Engineering, 21(3), pp.345–372.

Marinos, A. & Briscoe, G., 2009. Community cloud

computing. Lecture Notes in Computer Science

(including subseries Lecture Notes in Artificial

Intelligence and Lecture Notes in Bioinformatics), 5931

LNCS, pp.472–484.

Pahl, C., Xiong, H. & Walshe, R., 2013. A comparison of

on-premise to cloud migration approaches. Lecture

Notes in Computer Science (including subseries

Lecture Notes in Artificial Intelligence and Lecture

Notes in Bioinformatics), 8135 LNCS, pp.212–226.

Paraiso, F. et al., 2012. A federated multi-cloud PaaS

infrastructure. Proceedings - 2012 IEEE 5th

International Conference on Cloud Computing,

CLOUD 2012, pp.392–399.

Petrolo, R., Loscrí, V. & Mitton, N., 2014. Towards a smart

city based on cloud of things, a survey on the smart city

vision and paradigms. Proceedings of the 2014 ACM

international workshop on Wireless and mobile

technologies for smart cities - WiMobCity ’14, 25(3),

pp.61–66. Available at:

http://dl.acm.org/citation.cfm?id=2633661.2633667.

Rowe, F., Brinkley, J. & Tabrizi, N., 2013. Migrating

Legacy Applications to the Cloud. 2013 International

Conference on Cloud Computing and Big Data

(Cloudcom-Asia), (October 2009), pp.68–77.

Schmidt, F., MacDonell, S.G. & Connor, A.M., 2012. An

automatic architecture reconstruction and refactoring

framework. Studies in Computational Intelligence, 377,

pp.95–111.

Stal, M., 2007. Refactoring Software Architectures. In

Agile Software Architecture. Elsevier, pp. 63–82.

Available at:

http://linkinghub.elsevier.com/retrieve/pii/B97801240

77720000034.

Zimmermann, O., 2016. Architectural Refactoring for the

Cloud : a Decision-Centric View on Cloud Migration.

Computing.

Generic Refactoring Methodology for Cloud Migration - Position Paper

667