Microservices Adaptation using Machine Learning: A Systematic

Mapping Study

Anouar Hilali, Hatim Haﬁddi and Zineb El Akkaoui

InnovatiVE REsearch on Software, Systems and daTa,

Institut National des Postes et T

ecommunications, Rabat, Morocco

Keywords:

Self-adaptation, Microservices, Machine Learning.

Abstract:

The Microservice architecture is increasingly becoming the preferred architecture of modern applications.

The logically distinct components that make up microservices make continuous delivery easier compared

to monolithic architectures. This feature however makes it difﬁcult for engineers to control the underlying

services and properly adapt them at run-time. Designing our microservices as self-adaptive systems helps us

tackle this issue. Each microservice can then dynamically monitor and adapt its behavior to change certain

aspects of itself to achieve self-adaptive goals. The use of statistical and Machine Learning (ML) techniques

helps in this area in a lot of ways (e.g., predicting resource usage, anomaly detection, etc.). This paper aims to

provide a state of the art of the use of ML in microservice adaptation, the main goal is to provide an overview

of the ﬁeld and identify the most frequent adaptation goals and the types of adaptation techniques used. In

order to carry out a comprehensive analysis, a well-deﬁned method of systematic mapping is performed to

categorize, according to a detailed scheme, every paper relevant to this topic. The results can potentially shed

light on areas where further investigation might be warranted.

1 INTRODUCTION

Nowadays, modern applications are changing from

a monolithic architecture where the application is

constructed as a single entity, to a more distributed

architecture where complex applications are broken

down into logically recognizable components; mi-

croservices (Khazaei et al., 2018a). The Microser-

vice architecture is an architectural style in which

an application is designed as a set of small services

communicating with each other using a lightweight

mechanism (Fowler, 2014). The decoupling of com-

ponents that is characteristic of this type of architec-

ture makes continuous delivery safer and cheaper con-

trary to many other architectural styles (Junior, 2018).

However, several obstacles appear that make it dif-

ﬁcult to manually manage applications at run-time

(e.g., resource optimization, granularity of services,

context, fault detection, etc.). A solution to this is to

design these microservices as self-adaptive systems

that can dynamically monitor and adapt their behav-

ior to change certain aspects of themselves to meet

certain goals (e.g., when operating conditions are not

stable and optimal (Mendonc¸a et al., 2019), when ser-

vices need to be discovered in a changing context

(Wanigasekara, 2015a), etc.). Self-adaptation tech-

niques are known to be promising ways of tackling

and managing run-time uncertainties (Sanctis et al.,

2020a). We can distinguish four adaptation goals

in self-adaptive systems: self-conﬁguration (i.e., sys-

tems that conﬁgure themselves automatically), self-

optimization (systems that are constantly looking for

ways to optimize performance), self-healing (systems

that detect and ﬁx anomalies) and self-protection (sys-

tems that defend themselves from attacks) (Khazaei

et al., 2018a). An autonomic management system

contains the logic that tackles one or several of these

adaptation goals. The four elements: Monitor, Ana-

lyze, Plan, and Execute achieve the necessary func-

tions of any self-adaptive system. These elements

share common Knowledge therefore the model is usu-

ally referred to as the MAPE-K model (Kephart and

Chess, 2003). Applying statistical and ML techniques

in order to better any aspect of self-adaptive sys-

tems helps in bringing about adaptation without hu-

man intervention. ML algorithms for instance, can

be employed to predict resource usage patterns based

on historical data of an application’s microservices

and effectively anticipate changes (e.g., auto-scaling,

placement reconﬁguration, etc.) (Junior, 2018).

The goal of this paper is to achieve a research area

overview, identifying the most common microser-

vices adaptation goals, techniques that are used in

accomplishing these goals and the frequency of us-

ing ML algorithms in carrying this out. We there-

fore choose to perform a systematic mapping of the

Hilali, A., Haﬁddi, H. and El Akkaoui, Z.

Microservices Adaptation using Machine Learning: A Systematic Mapping Study.

DOI: 10.5220/0010578905210532

In Proceedings of the 16th International Conference on Software Technologies (ICSOFT 2021), pages 521-532

ISBN: 978-989-758-523-4

521

existent literature. Generally, a systematic mapping

study structures the type of research, reports and re-

sults that have been published by classifying them and

generates a visual summary of its results (Kitchenman

and Charters, 2007). In this work, the discussion of

the ﬁndings focus on supporting the understanding of

what has been addressed by the research community

and eventually propose future research guidelines in

this ﬁeld.

The remainder of this paper is structured as fol-

lows: section 2 details the research methodology used

for the mapping study accomplished in this work,

from establishing the research questions to the result-

ing systematic map. Then, in section 3 we provide

an analysis and a discussion of the results extracted

during the mapping study. In section 4 we discuss

the threats to validity of this paper. Section 5 spec-

iﬁes similar works and situates our contribution. Fi-

nally, section 6 outlines the main contribution of this

paper and considers some directions that could be po-

tentially investigated by the research community.

2 RESEARCH METHODOLOGY

In order to achieve the goals set above, our system-

atic mapping study process followed Kitchenman’s

approach. This type of study is helpful when it is dis-

covered that very little evidence is likely to exist on a

certain topic (Kitchenman and Charters, 2007). Fig-

ure 1 represents the underlying steps of performing a

systematic mapping study. Each step of the process

has an outcome, the last step produces the systematic

map. In the upcoming section we will describe each

step in detail and apply it to our particular case.

2.1 Deﬁnition of Research Questions

Initially we must deﬁne the research scope by identi-

fying goals that are expressed in research questions.

We chose to follow structured criteria to frame our

questions (Kitchenman and Charters, 2007), the fol-

lowing questions were stated:

RQ1: What are the microservices adaptation targets

(i.e., sub-goals of those mentioned in the introduc-

tion) that the research community tackles?

RQ2: Are the adaptations on an application or an ar-

chitecture level?

RQ3: What adaptation techniques were used?

Based on a classiﬁcation scheme that we detail later

on, we can answer more precise questions:

RQ4: If machine learning was used, what algorithms

were picked?

RQ5: What level of empirical evidence was attained

Definition of

Research

Question

Conduct Search

Screening of

Papers

Keywording

using Abstracts

Data Extraction

and Mapping

Process

Review Scope

All Papers

Relevant Papers

Classification

Scheme

Systematic Map

Process Steps Outcomes

Figure 1: Steps for performing a systematic mapping study.

by researchers?

2.2 Conduct Search for Primary Studies

In order to answer our questions, we must identify

the key studies by setting up a set of search strings

(Petersen et al., 2008). Multiple digital libraries were

chosen to extract relevant research using the two crite-

ria suggested by Kitchenman; Population (i.e. An ap-

plication area) and Intervention (a software method-

ology, tool, technology or procedure that addresses a

speciﬁc issue) (Kitchenman and Charters, 2007):

• Population: Microservices.

• Intervention: Adaptation, self-adaptation, self-

healing and context-aware.

Our Keywords can be deduced from each facet,

the population and intervention facets lead us to

a search string similar to this: (”microservices”

OR ”microservice” OR ”micro-service” OR ”micro-

services”) AND (”adaptation” OR ”adaptive” OR

”self-adaptation” OR ”self adaptation” OR ”self-

adaptive” OR ”self adaptive” OR ”context-aware”

OR ”context aware” OR ”self-healing” OR ”self

healing”).

The search query slightly varies from one digital

library to the other in order to comply with each li-

brary’s query rules.

The choice of what digital libraries to use was

based on (Breton et al., 2007)’s identiﬁcation of the

most relevant digital libraries to software engineers.

The choice of the search string was made in order

ICSOFT 2021 - 16th International Conference on Software Technologies

522

Table 1: Search queries used on different digital libraries

and their respective results.

Digital Library Number of results

ACM Digital Libray 16

IEEExplore 88

ScienceDirect 10

SpringerLink 149

Google Scholar 12

to select all papers that discuss the adaptation of mi-

croservices regardless of tools used (e.g. Artiﬁcial In-

telligence algorithms). We didn’t consider the com-

parison facet because it didn’t make sense in our case,

while the outcome facet would restrict our research

and wouldn’t allow us to paint a broad overview of

the research area. The results are accurate as of 10th

of January 2021, the number of total publications is

275. Table 1 shows the full results.

2.3 Screening of Papers for Inclusion

and Exclusion

Inclusion and exclusion criteria were used to only fo-

cus on papers that are relevant to answering our do-

main questions. A percentage of these criteria are

based on the fact that our main aim is microservice

architectures. Other criteria are based on practical is-

sues.

The inclusion criteria were:

IC1: publications produced in English.

IC2: publications that are peer-reviewed and pub-

lished in journals, conferences and workshops.

IC3: publications that explicitly mention our key-

words in the title, abstract and/or keywords.

These exclusion criteria were used:

EC1: publications where the authors focus on archi-

tectures other than microservices.

EC2: publications such as books, chapters of books

or similar ones.

EC3: publications that are not in the computer sci-

ence ﬁeld.

Duplicates are automatically dealt with using the

merge duplicates feature on the open source research

tool Zotero. Initially criteria IC1 was applied for all

queries. No ﬁlters were used for IEEE. Criteria IC2,

IC3, EC2 and EC3 were applied using the built-in

ﬁlters of the mentioned digital libraries. ”Research

Article” and Computer Science ﬁlters were applied

for ScienceDirect and ACM Digital Library and ﬁ-

nally ”Conference Paper” and ”Article” ﬁlters were

applied for SpringerLink. We then ﬁltered through

the publications manually, reading abstracts and if

necessary introductions and conclusions. We ranked

the papers based on their relevance to our objectives

by labeling each paper with a number (0 would be

irrelevant, 1 would be somewhat relevant and 2 as

in relevant). We double checked the somewhat rel-

evant papers a second time and discarded the irrel-

evant ones. Initially the number of relevant publi-

cations went from 275 to 129 based on reading ab-

stracts. Then the ﬁnal result was reduced to 62 pub-

lications that are useful for our purposes. Figure 2

illustrates this process.

2.4 Keywording of Relevant Papers

Before going ahead and setting up a classiﬁcation

scheme, the authors went through all the relevant pa-

pers and made sure they were all aligned with our ob-

jectives. We not only relied on reading the abstracts

but at times had to carefully read the full paper. In

order to properly cover all aspects of our research, the

papers were classiﬁed using the following classiﬁca-

tion scheme:

Adaptation Targets (RQ1). In order to answer RQ1,

we classiﬁed papers based on what area of microser-

vices were adapted.

Adaptation Techniques (RQ2). To classify the tech-

niques used in each research and address RQ2.

Adaptation Level (RQ3). To tackle RQ3, we also

classiﬁed papers based on what level of adaptation

were the techniques aimed at.

Use of Machine Learning (RQ4). To classify papers

on whether they used machine learning in their adap-

tations and if yes, what type of algorithms were used

in their solution.

Evidence (RQ5). We chose an existing classiﬁca-

tion of research approaches to classify our papers

(Wieringa et al., 2006) and address RQ5 on what type

of empirical evidence is provided by each paper:

• Validation Research: A paper where the tech-

niques that are examined are innovative and are

not yet applied in practice.

• Evaluation Research: A paper where the tech-

niques discussed were implemented in practice

and an evaluation of these techniques are con-

ducted.

• Solution Proposal: A paper where a solution to a

given problem is presented. It can be either new or

built on top of existing approaches. The solution

is also illustrated by a speciﬁc example.

• Philosophical Papers: Papers where a novel per-

spective is proposed by structuring a given ﬁeld.

• Opinion Papers: Papers where the author express

a personal opinion on the validity of a certain

technique or how it should be implemented.

Microservices Adaptation using Machine Learning: A Systematic Mapping Study

523

IEEE, ACM,

ScienceDirect,

SpringerLink,

Google Scholar

Reading title

and abstract

Reading introduction and

conclusion and/or full paper

Final set

275 129 62

Figure 2: Inclusion and exclusion criteria process.

• Experience Papers: Papers where an explanation

of how a technique has been done in practice is

given. It relies on the personal experience of the

authors.

2.5 Data Extraction and Mapping

Process

We conclude our review with the data extraction step.

All 64 papers relevant to us were fully read. Based

on the research questions deﬁned, an Excel data ex-

traction table was developed to document the data ex-

traction process. Thanks to the produced graphs, we

are now able to have a proper overview of what the re-

search community focuses on the microservices adap-

tation research area. Table 3 illustrates the detailed

scheme.

3 MAIN FINDINGS

All articles were published between September 2015

and January 2021. Figure 3 shows an increase in the

number of published articles on the the topic of Mi-

croservice adaptation. The decrease between 2020

and 2021 is due to the fact that the results were as

of the ﬁrst month of 2021.

Figure 3: The number of published papers per year since

2013.

3.1 RQ1: Adaptation Targets

Of all the texts that were reviewed, 42% (24 papers)

tackled Resource Scheduling or Utilization and Place-

ment, followed by discussions involving Auto-scaling

and Elasticity at 21.0% (17 papers), then Context at

11.1% (9 papers), Fault/Anomaly Detection and Ex-

ception Handling at 9.9% (8 papers) and Granular-

ity of Services at 3.7% (3 papers). Finally The Other

section with 12.3% (11 papers) constitutes solutions

tackling Orchestration/Collaboration, Task Work-

ﬂow/Scheduling, Data Transfer/Filtering/Acquisition,

Security/Fire-walling, Conﬁguration/Recovery, La-

tency and Network Resource Consumption. Figure

4 illustrates the distribution of goals. Table 2 details

each goal, its deﬁnition and the number of relevant

papers.

Figure 4: The distribution of the adaptation goals in relevant

papers.

We notice from the extracted data that there is a

huge focus on optimizing for performance. A com-

bined 63.0% between resource optimization and elas-

ticity. We believe this might be due to the fact that

industry professionals are mainly interested in opti-

mizing resources and researchers follow through by

proposing solutions for their concrete optimization is-

sues.

3.2 RQ2: Adaptation Level

Looking at the distribution of the adaptation level of

papers, we notice that the majority of papers (37 at

59.7%) present a solution acting exclusively on an ap-

plication level (i.e., solutions dealing with services,

containers and applications). Followed by papers (16

ICSOFT 2021 - 16th International Conference on Software Technologies

524

Table 2: The different deﬁnitions of adaptation goals and the number of papers discussing them.

Adaptation Target Deﬁnition Number of Papers

Resource Scheduling/Utilization,

Placement

Optimizing the use of resources

(e.g., CPU, memory, bandwidth, etc.)

and the adaptive placement of containers.

Auto-scaling and Elasticity

Automatically adjusting the capacity

to insure steady performance.

Context

The ability to collect data about

a system’s surrounding environment

(e.g., Location, Temperature, etc.).

Fault/Anomaly Detection,

Exception Handling

Detecting anomalies and errors in

the faulty microservices and

handling exceptions.

Granularity of Services

Identifying the optimal microservice

boundaries.

Orchestration/Collaboration

Automatic coordination of containers.

Task Workﬂow/Scheduling

Assigning tasks to the proper resources.

Data Transfer/Filtering/Acquisition

The ability to manipulate data.

Security/Fire-walling

Protection against malicious attacks.

Conﬁguration/Recovery

Having the proper system conﬁguration

(e.g., reliability, availability, performance, etc.)

and the ability to recover after failing.

Latency

The delay for data to go

from its source to its destination.

Network Resource Consumption

The underlying network’s resource

consumption.

at 25.8%) with both an application and architectural

(i.e., solutions dealing with how to design and archi-

tect your system) approach then purely architectural

solutions with 9 papers (14.5%). Figure 5 illustrates

this.

Figure 5: The distribution of the adaptation level in relevant

papers.

Since 53 papers (85.5%) have an application level

approach, we conclude that most research focuses on

solutions that tackle issues on the level of services, ap-

plications or containers. This makes sense since there

is a wide variety of possible contributions to have con-

trary to presenting solutions from an architectural and

design optic.

3.3 RQ3: Adaptation Techniques

From the present study, we believe that custom

techniques are the preferred approach for most re-

searchers. In addition, ML algorithms (i.e., classi-

cal machine learning and deep learning) are the tech-

niques of choice for numerous researchers in achiev-

ing adaptation targets. 10 papers employ a classical

machine learning algorithm in their solution. Rein-

forcement learning, Deep learning and heuristics are

used in 5 papers each. In 4 papers, researchers extend

the existing Kubernetes tools. A modiﬁed autoscaler

is introduced in 2 papers and so is the Non-Dominated

Sorting Genetic Algorithm (in order to make a con-

tainer allocation strategy and automatically manage

elasticity). The remaining papers (19) have the fol-

lowing techniques:

• Description language: A custom language and an

appropriate platform to aid in the implementation

of self-adaptive microservices,

• Resource scheduling algorithms: Policies used to

assign resources accordingly for energy efﬁciency

purposes,

Microservices Adaptation using Machine Learning: A Systematic Mapping Study

525

• External services: In-house developed software

and delegating run-time management to the cloud,

• Extended Berkeley Packet Filter: Used to inter-

cept key Linux system calls for efﬁcient monitor-

ing,

• Scaling policy derivation tool: Used to design and

evaluate 5 auto-scaling policies,

• Cloud-Edge-Dew architecture: A custom archi-

tecture that optimizes the advantages of Cloud and

Edge Computing to contribute microservices for

end user devices,

• Custom orchestrator: A custom orchestrating tool

for microservices,

• Horizontal ofﬂoading mechanism: A custom

mechanism based on the OpenFog reference ar-

chitecture where fog nodes can horizontally of-

ﬂoad computations to multiple less loaded nodes,

• Hierarchical scalable adaptive cloud monitoring

architecture: Custom architecture that monitors

physical and virtual infrastructure, realizes scal-

ability of monitoring based on microservices and

adjusts the monitoring interval and data transmis-

sion strategy (Wang et al., 2020).

• Latency aware microservice mashup algorithm:

A custom service mashup approach that focuses

on network resource consumption in edge net-

work in addition to efﬁcient service mashup,

• Micro-controllers: A controller that can be com-

posed or conﬁgured at run-time based on the

adaptation goals of the target system,

• Model based approach: A custom method called

Microservices Recovery Action Selection that

adapts to frequent microservices changes without

the need for historical data of previous failures,

• Microservice ambients using aspects: A model-

ing concept that considers microservice bound-

aries as an adaptable ﬁrst-class entity, it is also

based on the aspect-oriented architectural meta-

modeling approach of ambients (Hassan et al.,

2017),

• Architecture and planning engine at run-time: An

interactive and iterative planning engine that pro-

vides insight into which granularity adaptation

strategy is adequate at run-time (Hassan, 2019),

• Custom framework: A self-adaptive root cause

diagnosis framework in order to analyze several

metrics collected from given microservices,

• Event processing techniques: Used as an inte-

grated part to an appropriate architecture to better

smart decision-making,

• Application partitioning and task ofﬂoading algo-

rithm: Which comes up with an application parti-

tioning decision at run-time,

• Application-infrastructure co-programming

model and architecture: Which provides a

controllable environment for the creation of

application logic and enable reconﬁgurability of

computing resources at run-time according to the

needs of a speciﬁc application (

Stefani

c et al.,

2019) and

• Vehicular-OBUs-As-On-Demand-Fogs: A custom

framework that efﬁciently uses clusters of vehi-

cles to beneﬁt from on-board units in order to op-

timize resources.

3.4 RQ4: Use of Machine Learning

Here we notice that although small, there is a decent

interest in using Artiﬁcial Intelligence in order to bet-

ter microservice adaptation. 40.3% (25 papers) of pa-

pers utilize either classical machine learning or deep

learning algorithms, the remaining 37 papers (59.7%)

use other techniques mentioned in section 3.3. The

following algorithms were used:

• Data Classiﬁcation algorithms,

• Support Vector Machine,

• Bayesian Network,

• Decision Tree,

• Naive Bayes classiﬁer,

• Contextual Bandit Reinforcement Learning,

• Stacked Long Short-Term Memory Networks,

• K-means,

• Planning algorithms,

• Extreme Learning Machine,

• Deep Q-Learning,

• Fuzzy Lattice Reasoning,

• Artiﬁcial Neural Network and

• Gaussian Process Regression.

Figure 6 illustrates the distribution of what targets

researchers go to when utilizing ML as an adaptation

technique. We notice that resource optimization is

predominately the most common one with 53.1% of

total papers that use ML.

3.5 RQ5: Evidence

The level of empirical evidence in the relevant pa-

pers is dominated by solution proposals at 54 papers

ICSOFT 2021 - 16th International Conference on Software Technologies

526

Figure 6: Distribution of the most commonly tackled targets

when using ML.

Figure 7: The distribution of types of papers.

(87.1%), followed by 4 (6.5%) opinion papers, 2 eval-

uation papers and 2 validation research papers (3.2%).

Figure 7 shows the distribution.

This means that the research community is mainly

focused on solving concrete microservice adaptation

problems and supporting their solutions with real ex-

amples. The small number of papers (6 at 9.7%)

where practical applications are not provided (i.e.,

validation research, philosophical, opinion and expe-

rience papers) indicates that there is virtually no to

little to no interest in discussions that do not present a

valid practical example.

4 THREATS TO VALIDITY

Assessing Threats to Validity is clinical in order to en-

sure quality empirical studies in Software Engineer-

ing (Zhou et al., 2016).

Construct Validity: The discussion surrounding our

main ﬁndings in section 3 is only valid for the pa-

pers provided. We therefore made sure to include all

the possible relevant papers. In order to realize this,

we used all the databases that are relevant to software

engineering research mentioned in section 2.2. We

made sure to answer the appropriate research ques-

tions and deduce the adequate complete search terms,

we for instance considered all the possible variations

of our facets (e.g., microserices, self-adaptation). We

also avoided the threat of inappropriate inclusion and

exclusion criteria in our screening by considering the

tile, abstract and keywords. In addition, we decided

as an initial safety measure not to exclude papers that

haven’t been fully investigated until there has been an

exhaustive reading.

Internal Validity: In order to alleviate the issues sur-

rounding data extraction and classiﬁcation, we took

the proper time-frame to, at times, read introductions

and conclusion and some times the entirety of a given

paper if it wasn’t possible to extract data initially from

the abstract. We also used an Excel table to properly

store the extracted data to generate relevant statistics:

External Validity: We made sure not to restrict the

time span of the resulting studies. To improve exter-

nal validity in the future it is perhaps advised to have

full access to all relevant papers by contacting their

respective authors.

Conclusion Validity: The study’s replicability is pos-

sible thanks to the search method details provided in

section 2. The threat of primary study duplication was

also avoided by using the open search tool Zotero and

always double checking our data.

5 RELATED WORK

Numerous systematic reviews tackled self-adaptation.

One paper outlines the machine learning techniques

used to approach self adaptation based on the con-

cerns, aspects and purposes of choice (Saputri and

Lee, 2020), although exhaustive we believe it wasn’t

applied to a speciﬁc type of architecture. Multiple

papers discuss self-adaptation through the lens of a

speciﬁc context (e.g., mobile app (Grua et al., 2019),

cyber-physical system (Muccini et al., 2016), etc.).

Although there is a discussion of self-adaptation in

Service Oriented Architecture (Romay et al., 2011)

there are no equivalent for the Microservices Archi-

tecture. To the best of our knowledge there were no

Systematic Mapping Reviews discussing the use of

ML techniques for the adaptation of microservices.

6 CONCLUSIONS

In this paper, we provide the results concerning the

systematic mapping study of the use of ML in Mi-

croservice adaptation. this paper outlines important

observations:

• recognizing the most focused on microservices

Microservices Adaptation using Machine Learning: A Systematic Mapping Study

527

adaptation targets in the research community;

• highlighting the techniques of choice that authors

go to;

• ﬁguring out the level of empirical evidence

achieved by the researchers’ presented solutions;

Our study, with the help of the developed classi-

ﬁcation scheme, provides the most commonly tack-

led issues in microservice adaptation targets; resource

optimization and elasticity. It also indicates that

although many authors decide to develop a unique

adaptation technique, there is also a growing inter-

est in using ML algorithms as techniques to achieve

adaptation targets. Our ﬁndings also reﬂect that the

majority of researchers provide practical solutions to

concrete issues.

Our objective was to provide a comprehensive

mapping of the use of ML in microservices adap-

tation. We intend to go more in depth and specify

the strengths and weaknesses of speciﬁc techniques in

speciﬁc contexts, explore opportunities where adapta-

tion techniques can intersect and be more efﬁcient at

general self-adaptation and ﬁnally add to the growing

discussion by introducing concepts that might help us

make abstractions to adaptation targets and build self-

adaptive microservices that comprise of all the main

goals (i.e., self conﬁguration, self-optimization, self-

healing and self-protection), which will be our future

work.

REFERENCES

Abdullah, M., Iqbal, W., Bukhari, F., and Erradi, A. (2020).

Diminishing Returns and Deep Learning for Adap-

tive CPU Resource Allocation of Containers. IEEE

Transactions on Network and Service Management,

17(4):2052–2063. Conference Name: IEEE Transac-

tions on Network and Service Management.

Barna, C., Khazaei, H., Fokaefs, M., and Litoiu, M. (2017).

Delivering Elastic Containerized Cloud Applications

to Enable DevOps. In 2017 IEEE/ACM 12th Interna-

tional Symposium on Software Engineering for Adap-

tive and Self-Managing Systems (SEAMS), pages 65–

75.

Bellur, U., Narendra, N. C., and Mohalik, S. K. (2017). AU-

SOM: Autonomic Service-Oriented Middleware for

IoT-Based Systems. In 2017 IEEE World Congress

on Services (SERVICES), pages 102–105.

Breton, P., Kitchenman, B., Budgen, D., Turner, M., and

Khalil, M. (2007). Lessons from applying the system-

atic literature review process within the software en-

gineering domain. j syst softw. In Journal of Systems

and Software.

Casalicchio, E. (2019). A study on performance measures

for auto-scaling CPU-intensive containerized applica-

tions. Cluster Comput, 22(3):995–1006.

Chegini, H. and Mahanti, A. (2019). A Framework of Au-

tomation on Context-Aware Internet of Things (IoT)

Systems. In Proceedings of the 12th IEEE/ACM Inter-

national Conference on Utility and Cloud Computing

Companion, UCC ’19 Companion, pages 157–162,

New York, NY, USA. Association for Computing Ma-

chinery.

Coulson, N. C., Sotiriadis, S., and Bessis, N. (2020). Adap-

tive Microservice Scaling for Elastic Applications.

IEEE Internet of Things Journal, 7(5):4195–4202.

Conference Name: IEEE Internet of Things Journal.

De Sanctis, M., Bucchiarone, A., and Marconi, A. (2020).

Dynamic adaptation of service-based applications: a

design for adaptation approach. J Internet Serv Appl,

11(1):2.

Donca, I., Corches, C., Stan, O., and Miclea, L. (2020).

Autoscaled RabbitMQ Kubernetes Cluster on single-

board computers. In 2020 IEEE International Con-

ference on Automation, Quality and Testing, Robotics

(AQTR), pages 1–6.

Florio, L. and Nitto, E. D. (2016). Gru: An Approach to

Introduce Decentralized Autonomic Behavior in Mi-

croservices Architectures. In 2016 IEEE International

Conference on Autonomic Computing (ICAC), pages

357–362.

Fowler, M. (2014). Microservices.

Grua, E. M., Malavolta, I., and Lago, P. (2019). Self-

adaptation in mobile apps: a systematic literature

study. In 2019 IEEE/ACM 14th International Sympo-

sium on Software Engineering for Adaptive and Self-

Managing Systems (SEAMS).

Guerrero, C., Lera, I., and Juiz, C. (2018a). Genetic Algo-

rithm for Multi-Objective Optimization of Container

Allocation in Cloud Architecture. J Grid Computing,

16(1):113–135.

Guerrero, C., Lera, I., and Juiz, C. (2018b). Resource op-

timization of container orchestration: a case study in

multi-cloud microservices-based applications. J Su-

percomput, 74(7):2956–2983.

Hassan, S. (2019). Modelling and evaluation of microser-

vice granularity adaptation decisions. PhD Thesis,

University of Birmingham.

Hassan, S., Ali, N., and Bahsoon, R. (2017). Microser-

vice Ambients: An Architectural Meta-Modelling Ap-

proach for Microservice Granularity. In 2017 IEEE

International Conference on Software Architecture

(ICSA), pages 1–10.

Hassan, S. and Bahsoon, R. (2016). Microservices and

Their Design Trade-Offs: A Self-Adaptive Roadmap.

In 2016 IEEE International Conference on Services

Computing (SCC), pages 813–818.

Herrera, J. and Molt

o, G. (2020). Toward Bio-Inspired

Auto-Scaling Algorithms: An Elasticity Approach

for Container Orchestration Platforms. IEEE Access,

8:52139–52150. Conference Name: IEEE Access.

Houmani, Z., Balouek-Thomert, D., Caron, E., and

Parashar, M. (2020). Enhancing microservices archi-

tectures using data-driven service discovery and QoS

guarantees. In 2020 20th IEEE/ACM International

ICSOFT 2021 - 16th International Conference on Software Technologies

528

Symposium on Cluster, Cloud and Internet Computing

(CCGRID), pages 290–299.

Imdoukh, M., Ahmad, I., and Alfailakawi, M. G.

(2020). Machine learning-based auto-scaling for con-

tainerized applications. Neural Comput & Applic,

32(13):9745–9760.

Jim

enez, L. L. and Schel

en, O. (2019). DOCMA: A Decen-

tralized Orchestrator for Containerized Microservice

Applications. In 2019 IEEE Cloud Summit, pages 45–

51.

Junior, A. (2018). Runtime adaptation of microservices.

Kang, P. and Lama, P. (2020). Robust Resource Scaling of

Containerized Microservices with Probabilistic Ma-

chine learning. In 2020 IEEE/ACM 13th International

Conference on Utility and Cloud Computing (UCC),

pages 122–131.

Keni, N. D. and Kak, A. (2020). Adaptive Containerization

for Microservices in Distributed Cloud Systems. In

2020 IEEE 17th Annual Consumer Communications

Networking Conference (CCNC), pages 1–6. ISSN:

2331-9860.

Kephart, J. and Chess, D. (2003). The vision of autonomic

computing. IEEE.

Khazaei, H., Ghanbari, A., and Litoiu, M. (2018a). Adap-

tation as a service. In Proceedings of the 28th An-

nual International Conference on Computer Science

and Software Engineering. IBM Corp.

Khazaei, H., Ghanbari, A., and Litoiu, M. (2018b). Adap-

tation as a service. In Proceedings of the 28th An-

nual International Conference on Computer Science

and Software Engineering, CASCON ’18, pages 282–

288, USA. IBM Corp.

Kitchenman, B. and Charters, S. (2007). Guidelines for per-

forming systematic literature reviews in software en-

gineering.

Klinaku, F., Frank, M., and Becker, S. (2018). CAUS:

An Elasticity Controller for a Containerized Microser-

vice. In Companion of the 2018 ACM/SPEC Interna-

tional Conference on Performance Engineering, ICPE

’18, pages 93–98, New York, NY, USA. Association

for Computing Machinery.

Kumar, J. and Singh, A. K. (2020). Decomposition

Based Cloud Resource Demand Prediction Using Ex-

treme Learning Machines. J Netw Syst Manage,

28(4):1775–1793.

Lakhan, A. and Li, X. (2020). Transient fault aware ap-

plication partitioning computational ofﬂoading algo-

rithm in microservices based mobile cloudlet net-

works. Computing, 102(1):105–139.

Ma, M., Lin, W., Pan, D., and Wang, P. (2019). MS-Rank:

Multi-Metric and Self-Adaptive Root Cause Diagno-

sis for Microservice Applications. In 2019 IEEE Inter-

national Conference on Web Services (ICWS), pages

60–67.

Magableh, B. (2016). Deep Q Learning for Self Adap-

tive Distributed Microservices Architecture (in press).

4:17.

Meixner, S., Schall, D., Li, F., Karagiannis, V., Schulte,

S., and Plakidas, K. (2019). Automatic Application

Placement and Adaptation in Cloud-Edge Environ-

ments. In 2019 24th IEEE International Conference

on Emerging Technologies and Factory Automation

(ETFA), pages 1001–1008. ISSN: 1946-0759.

Mendonc¸a, N., Jamshidi, P., Garlan, D., and Pahl, C.

(2019). Developing self-adaptive microservice sys-

tems: Challenges and directions. In IEEE Software

2019. IEEE.

Mostafa, M. A. A. and Khater, A. M. (2019). Horizon-

tal Ofﬂoading Mechanism for IoT Application in Fog

Computing Using Microservices Case Study: Trafﬁc

Management System. In 2019 IEEE Jordan Interna-

tional Joint Conference on Electrical Engineering and

Information Technology (JEEIT), pages 640–647.

Muccini, H., Sharaf, M., and Weyns, D. (2016). Self-

adaptation for cyber-physical systems: a systematic

literature review. In Proceedings of the 11th In-

ternational Symposium on Software Engineering for

Adaptive and Self-Managing Systems. Association for

Computing Machinery.

Nabi, S. and Ahmed, M. (2021). OG-RADL: over-

all performance-based resource-aware dynamic load-

balancer for deadline constrained Cloud tasks. J Su-

percomput.

Neves, F., Vilac¸a, R., and Pereira, J. (2020). Black-box

inter-application trafﬁc monitoring for adaptive con-

tainer placement. In Proceedings of the 35th Annual

ACM Symposium on Applied Computing, SAC ’20,

pages 259–266, New York, NY, USA. Association for

Computing Machinery.

Nguyen, P. and Nahrstedt, K. (2017). MONAD: Self-

Adaptive Micro-Service Infrastructure for Heteroge-

neous Scientiﬁc Workﬂows. In 2017 IEEE Interna-

tional Conference on Autonomic Computing (ICAC),

pages 187–196. ISSN: 2474-0756.

Orsini, G., Posdorfer, W., and Lamersdorf, W. (2019). Efﬁ-

cient Mobile Clouds: Forecasting the Future Connec-

tivity of Mobile and IoT Devices to Save Energy and

Bandwidth. Procedia Computer Science, 155:121–

128.

Ortiz, G., Caravaca, J. A., Garc

ıa-de Prado, A., O, F. C.

d. l., and Boubeta-Puig, J. (2019). Real-Time Context-

Aware Microservice Architecture for Predictive An-

alytics and Smart Decision-Making. IEEE Access,

7:183177–183194. Conference Name: IEEE Access.

Pahl, M. and Aubet, F. (2018). All Eyes on You: Distributed

Multi-Dimensional IoT Microservice Anomaly De-

tection. In 2018 14th International Conference on

Network and Service Management (CNSM), pages

72–80. ISSN: 2165-963X.

Petersen, K., Feldt, R., Mujtaba, S., and Mattsson, M.

(2008). Systematic mapping studies in software engi-

neering. In EASE’08: Proceedings of the 12th Inter-

national Conference on Evaluation and Assessment in

Software Engineering, Swindon, GBR. BCS Learning

& Development Ltd.

Podolskiy, V., Jindal, A., Gerndt, M., and Oleynik, Y.

(2018). Forecasting Models for Self-Adaptive Cloud

Applications: A Comparative Study. In 2018 IEEE

12th International Conference on Self-Adaptive and

Microservices Adaptation using Machine Learning: A Systematic Mapping Study

529

Self-Organizing Systems (SASO), pages 40–49. ISSN:

1949-3681.

Ramirez, Y. M., Podolskiy, V., and Gerndt, M. (2019).

Capacity-Driven Scaling Schedules Derivation for

Coordinated Elasticity of Containers and Virtual Ma-

chines. In 2019 IEEE International Conference on

Autonomic Computing (ICAC), pages 177–186. ISSN:

2474-0756.

Ravandi, B. and Papapanagiotou, I. (2018). A self-

organized resource provisioning for cloud block stor-

age. Future Generation Computer Systems, 89:765–

776.

Rodr

ıguez-Gracia, D., Piedra-Fern

andez, J. A., Iribarne,

L., Criado, J., Ayala, R., Alonso-Montesinos, J., and

Maria de las Mercedes, C.-U. (2019). Microservices

and Machine Learning Algorithms for Adaptive Green

Buildings. Sustainability, 11(16):4320. Publisher:

Multidisciplinary Digital Publishing Institute.

Romay, M. P., Fernandez-Sanz, L., and Rodriguez, D.

(2011). A systematic review of self-adaptation in

service-oriented architectures.

Rossi, F., Cardellini, V., and Presti, F. L. (2020a). Hierarchi-

cal Scaling of Microservices in Kubernetes. In 2020

IEEE International Conference on Autonomic Com-

puting and Self-Organizing Systems (ACSOS), pages

28–37.

Rossi, F., Cardellini, V., and Presti, F. L. (2020b). Self-

adaptive Threshold-based Policy for Microservices

Elasticity. In 2020 28th International Symposium on

Modeling, Analysis, and Simulation of Computer and

Telecommunication Systems (MASCOTS), pages 1–8.

ISSN: 2375-0227.

Rovnyagin, M. M., Guminskaia, A. V., Plyukhin, A. A.,

Orlov, A. P., Chernilin, F. N., and Hrapov, A. S.

(2018). Using the ML-based architecture for adaptive

containerized system creation. In 2018 IEEE Confer-

ence of Russian Young Researchers in Electrical and

Electronic Engineering (EIConRus), pages 358–360.

Rychener, L., Montet, F., and Hennebert, J. (2020). Archi-

tecture Proposal for Machine Learning Based Indus-

trial Process Monitoring. Procedia Computer Science,

170:648–655.

Sahni, J. and Vidyarthi, D. P. (2017). Heterogeneity-aware

adaptive auto-scaling heuristic for improved QoS and

resource usage in cloud environments. Computing,

99(4):351–381.

Sami, H., Mourad, A., and El-Hajj, W. (2020). Vehicular-

OBUs-As-On-Demand-Fogs: Resource and Context

Aware Deployment of Containerized Micro-Services.

IEEE/ACM Transactions on Networking, 28(2):778–

790. Conference Name: IEEE/ACM Transactions on

Networking.

Sampaio, A. R., Rubin, J., Beschastnikh, I., and Rosa, N. S.

(2019). Improving microservice-based applications

with runtime placement adaptation. J Internet Serv

Appl, 10(1):4.

Sanctis, M., Muccini, H., and Vaidhyanathan, K. (2020a).

Data-driven adaptation in microservice-based iot ar-

chitectures. In 2020 IEEE International Conference

on Software Architecture Companion (ICSA-C).

Sanctis, M. D., Muccini, H., and Vaidhyanathan, K.

(2020b). Data-driven Adaptation in Microservice-

based IoT Architectures. In 2020 IEEE Interna-

tional Conference on Software Architecture Compan-

ion (ICSA-C), pages 59–62.

Saputri, T. R. D. and Lee, S.-W. (2020). The application of

machine learning in self-adaptive systems: A system-

atic literature review. In IEEE Access. IEEE Access.

Siqueira, B. R., Ferrari, F. C., Vogel, T., and Lemos, R. d.

(2020). Micro-controllers: Promoting Structurally

Flexible Controllers in Self-Aware Computing Sys-

tems. In 2020 IEEE International Conference on

Autonomic Computing and Self-Organizing Systems

Companion (ACSOS-C), pages 188–193.

Stefani

c, P., Cigale, M., Jones, A. C., Knight, L., Taylor,

I., Istrate, C., Suciu, G., Ulisses, A., Stankovski, V.,

Taherizadeh, S., Salado, G. F., Koulouzis, S., Mar-

tin, P., and Zhao, Z. (2019). SWITCH workbench:

A novel approach for the development and deploy-

ment of time-critical microservice-based cloud-native

applications. Future Generation Computer Systems,

99:197–212.

Tefera, G., She, K., and Deeba, F. (2019). Decentral-

ized Adaptive Latency-Aware Cloud-Edge-Dew Ar-

chitecture for Unreliable Network. In Proceedings of

the 2019 11th International Conference on Machine

Learning and Computing, ICMLC ’19, pages 142–

146, New York, NY, USA. Association for Computing

Machinery.

Wang, R., Ying, S., Li, M., and Jia, S. (2020). HSACMA: a

hierarchical scalable adaptive cloud monitoring archi-

tecture. Software Qual J, 28(3):1379–1410.

Wang, W., Fan, L., Huang, P., and Li, H. (2019). A New

Data Processing Architecture for Multi-Scenario Ap-

plications in Aviation Manufacturing. IEEE Access,

7:83637–83650. Conference Name: IEEE Access.

Wang, X., Feng, Z., and Huang, K. (2018). D3L-Based

Service Runtime Self-Adaptation Using Replanning.

IEEE Access, 6:14974–14995. Conference Name:

IEEE Access.

Wang, Y. (2019). Towards service discovery and autonomic

version management in self-healing microservices ar-

chitecture. In Proceedings of the 13th European Con-

ference on Software Architecture - Volume 2, ECSA

’19, pages 63–66, New York, NY, USA. Association

for Computing Machinery.

Wanigasekara, N. (2015a). A semi lazy bandit approach

for intelligent service discovery in iot applications. In

Adjunct Proceedings of the 2015 ACM International

Joint Conference on Pervasive and Ubiquitous Com-

puting and Proceedings of the 2015 ACM Interna-

tional Symposium on Wearable Computers. Associa-

tion for Computing Machinery.

Wanigasekara, N. (2015b). A semi lazy bandit approach

for intelligent service discovery in IoT applications.

In Adjunct Proceedings of the 2015 ACM Interna-

tional Joint Conference on Pervasive and Ubiquitous

Computing and Proceedings of the 2015 ACM Inter-

national Symposium on Wearable Computers, Ubi-

Comp/ISWC’15 Adjunct, pages 503–508, New York,

NY, USA. Association for Computing Machinery.

ICSOFT 2021 - 16th International Conference on Software Technologies

530

Wieringa, R., Maiden, N., Mead, N., and Rolland, C.

(2006). Requirements engineering paper classiﬁcation

and evaluation criteria: A proposal and a discussion.

In Requir. Eng.

Wu, L., Tordsson, J., Acker, A., and Kao, O. (2020). Micro-

RAS: Automatic Recovery in the Absence of Histor-

ical Failure Data for Microservice Systems. In 2020

IEEE/ACM 13th International Conference on Utility

and Cloud Computing (UCC), pages 227–236.

Xu, M., Toosi, A. N., and Buyya, R. (2020). A Self-adaptive

Approach for Managing Applications and Harnessing

Renewable Energy for Sustainable Cloud Computing.

IEEE Transactions on Sustainable Computing, pages

1–1. Conference Name: IEEE Transactions on Sus-

tainable Computing.

Yang, Z., Nguyen, P., Jin, H., and Nahrstedt, K. (2019). MI-

RAS: Model-based Reinforcement Learning for Mi-

croservice Resource Allocation over Scientiﬁc Work-

ﬂows. In 2019 IEEE 39th International Conference on

Distributed Computing Systems (ICDCS), pages 122–

132. ISSN: 2575-8411.

Yudong, L., Yuqing, Z., and Zhangbin, Z. (2020). Service

Availability Guarantee with Adaptive Automatic Flow

Control. In 2020 IEEE World Congress on Services

(SERVICES), pages 101–105. ISSN: 2642-939X.

Zang, X., Chen, W., Zou, J., Zhou, S., Lisong, H., and

Ruigang, L. (2018). A Fault Diagnosis Method for

Microservices Based on Multi-Factor Self-Adaptive

Heartbeat Detection Algorithm. In 2018 2nd IEEE

Conference on Energy Internet and Energy System In-

tegration (EI2), pages 1–6.

Zhang, S., Zhang, M., Ni, L., and Liu, P. (2019). A

Multi-Level Self-Adaptation Approach For Microser-

vice Systems. In 2019 IEEE 4th International Con-

ference on Cloud Computing and Big Data Analysis

(ICCCBDA), pages 498–502.

Zheng, T., Zheng, X., Zhang, Y., Deng, Y., Dong, E., Zhang,

R., and Liu, X. (2019). SmartVM: a SLA-aware mi-

croservice deployment framework. World Wide Web,

22(1):275–293.

Zhou, A., Wang, S., Wan, S., and Qi, L. (2020). LMM:

latency-aware micro-service mashup in mobile edge

computing environment. Neural Comput & Applic,

32(19):15411–15425.

Zhou, X., Jin, Y., Zhang, H., Li, S., and Huang, X. (2016).

A map of threats to validity of systematic literature

reviews in software engineering. In 2016 23rd Asia-

Paciﬁc Software Engineering Conference (APSEC).

IEEE.

APPENDIX

As mentioned in section 2.5, table 3 represents the

detailed scheme we used to conduct our SMR.

Microservices Adaptation using Machine Learning: A Systematic Mapping Study

531

Table 3: The Systematic Mapping Review’s detailed scheme.

Number Paper Year Target Technique Evidence

Use of

Architecture

level

Application

level

1 (Zang et al., 2018) 2018 Fault detection

Multi-factor self-adaptive

heartbeat detection algorithm

Solution

proposal

No No Yes

2 (Chegini and Mahanti, 2019) 2019

Context, orchestration

and collaboration, task workﬂow

and scheduling, data transfer and

ﬁltering

Knowledge or rule-based component,

Data classiﬁcation algorithms (ML) and

data science and feature engineering

Evaluation

paper

Yes No Yes

3 (Zhang et al., 2019) 2019 Service and instance numbers

Microservice self-adaptation description

language and adaptive K8 (Extending Kubernetes)

Solution

proposal

No No Yes

4 (Wang et al., 2019) 2019 Data acquisition Big Data processing

Solution

proposal

No Yes No

5 (Xu et al., 2020) 2020 Resource scheduling

Resource scheduling algorithms, SVM to predict

solar irradiation or PV power output for the availability

of renewable energy

Solution

proposal

Yes No Yes

6 (Ravandi and Papapanagiotou, 2018) 2018 Resource utilization

Bayesian network, decision tree, Naive bayes

and Support Vector Machine

Solution

proposal

Yes No Yes

7 (Wanigasekara, 2015b) 2015 Context Contextual Bandit Reinforcement Learning algorithms

Solution

proposal

Yes No Yes

8 (Casalicchio, 2019) 2019 Resource utilization, scaling

KHPA-A algorithm (Extending Kubernetes’s autoscaling

algorithm based on CPU usage)

Solution

proposal

No No Yes

9 (Khazaei et al., 2018b) 2018

Security, conﬁguration,

fault detection and resource

optimization

Adaptation as external services

Solution

proposal

Yes Yes No

10 (Keni and Kak, 2020) 2020 SLA based resource allocation

Optimal containerization framework using

mathematical representations

Solution

proposal

No No Yes

11 (Coulson et al., 2020) 2020

Resource allocation

and auto-scaling

Stacked Long Short-Term Memory Network

Solution

proposal

Yes No Yes

12 (Pahl and Aubet, 2018) 2018

Anomaly detection

and ﬁre-walling

Grid-based algorithm, k-mea,s

Solution

proposal

Yes No Yes

13 (Rychener et al., 2020) 2020 Anomaly detetion Machine learning

Opinion

paper

Yes Yes Yes

14 (Bellur et al., 2017) 2017

Context, application

requirement

Service-oriented middleware

Solution

paper

No Yes Yes

15 (Meixner et al., 2019) 2019 Resource and placement

NFRs based on user deﬁned rules for placement

and data-driven automatic deployment

Solution

proposal

Yes Yes No

16 (Donca et al., 2020) 2020 Resource and auto-scaling Kubernetes auto-scaler and Raspberry Pi

Solution

proposal

No Yes No

17 (Neves et al., 2020) 2020 Adaptive placement Extended Berkeley Packet Filter

Solution

proposal

No No Yes

18 (Ramirez et al., 2019) 2019 Auto-scaling Scaling Policy Derivation Tool

Solution

proposal

No No Yes

19 (Klinaku et al., 2018) 2018 Resource and auto-scaling Heuristics

Solution

proposal

No No Yes

20 (Wang et al., 2018) 2018 Exception handling AI Planning algorithms

Solution

proposal

Yes No Yes

21 (Sanctis et al., 2020b) 2020 Resource Machine Learning, Q-learning, AI Planning

Validation

research

Yes Yes Yes

22 (Tefera et al., 2019) 2019 Latency Cloud-Edge-Dew Architecture

Validation

Research

No Yes No

23 (Kumar and Singh, 2020) 2020 Resource Extreme Learning Machine

Solution

proposal

Yes No Yes

24 (Magableh, 2016) 2016 Anomaly detection

Deep Q Learning,

Reinforcement Learning

Solution

proposal

Yes Yes Yes

25 (Barna et al., 2017) 2017 Auto-scaling

Self-tuning performance model

and custom autonomic management system

Solution

proposal

No Yes No

26 (Abdullah et al., 2020) 2020 Resource Deep Learning

Solution

proposal

Yes No Yes

27 (Jim

enez and Schel

en, 2019) 2019 Resource Custom orchestrator; DOCMA

Solution

proposal

No Yes Yes

28 (De Sanctis et al., 2020) 2020 Resource Architecture

Solution

proposal

Yes Yes No

29 (Orsini et al., 2019) 2019 Context Machine Learning

Solution

proposal

Yes No Yes

30 (Houmani et al., 2020) 2020 Resource

Custom auto-scaler, scale-up/down and

load shedding algorithm

Solution

proposal

No Yes Yes

31 (Podolskiy et al., 2018) 2018 Auto-scaling, resource

Machine Learning, ARIMA,

GARCH SSA, SVR

Solution

proposal

Yes No Yes

32 (Guerrero et al., 2018a) 2018

Resource, container allocation

and elasticity management

Non-dominated Sorting Genetic Algorithm II

Solution

proposal

No No Yes

33 (Florio and Nitto, 2016) 2016 Auto-scaling, resource Autonomic manager, MAPE-K

Solution

proposal

No No Yes

34 (Sahni and Vidyarthi, 2017) 2017 Auto-scaling Heuristics

Solution

proposal

No No Yes

35 (Rossi et al., 2020a) 2020 Elasticity, Auto-scaling

Custom Auto-scaler, Kubernetes extension,

RL based scaling

Solution

proposal

Yes No Yes

36 (Mostafa and Khater, 2019) 2019 Context Horizontal ofﬂoading mechanism

Solution

proposal

No Yes Yes

37 (Wang et al., 2020) 2020 Resource

HSACMA (Hierarchical Scalable Adaptive

Cloud Monitoring Architecture)

Solution

proposal

No Yes Yes

38 (Sampaio et al., 2019) 2019 Resource, placement REMap (custom adaptation mechanism)

Solution

proposal

No No Yes

39 (Zhou et al., 2020) 2020 Network resource consumption Latency aware microservice mashup algorithm

Solution

proposal

No No Yes

40 (Imdoukh et al., 2020) 2020 Auto-scaling, resource LSTM on different data types

Solution

proposal

Yes No Yes

41 (Siqueira et al., 2020) 2020 Context Micro-controllers

Solution

proposal

No No Yes

42 (Wu et al., 2020) 2020 Recovery Model based approach

Solution

proposal

No No Yes

43 (Hassan et al., 2017) 2017 Granularity of services MS Ambients using Aspects

Solution

proposal

No Yes No

44 (Rodr

ıguez-Gracia et al., 2019) 2019 Context, energy consumption

ML for decision making, Fuzzy

Lattice Reasoning (FLR)

Solution

proposal

Yes Yes Yes

45 (Hassan and Bahsoon, 2016) 2016 Granularity of services Modiﬁed MAPE-K

Opinion

paper

No No Yes

46 (Yang et al., 2019) 2019 Resource Reinforcement Learning

Solution

proposal

Yes No Yes

47 (Hassan, 2019) 2019 Granularity of services Architecture and planning engine for insight at runtime

Solution

proposal

No Yes Yes

48 (Nguyen and Nahrstedt, 2017) 2017 Resource, scheduling Artiﬁcial Neural Network for system identiﬁcation

Solution

proposal

Yes Yes No

49 (Ma et al., 2019) 2019 Anomaly detection Custom framework, algorithm

Solution

proposal

No No Yes

50 (Nabi and Ahmed, 2021) 2021 Resource and Task scheduling Heuristics

Solution

proposal

No No Yes

51 (Ortiz et al., 2019) 2019 Context Event processing techniques

Solution

proposal

No Yes No

52 (Guerrero et al., 2018b) 2018 Resource Non-dominated Sorting Genetic Algorithm

Evaluation

paper

No No Yes

53 (Kang and Lama, 2020) 2020 Resource

Probabilistic Machine Learning based models;

Gaussian Process Regression

Solution

proposal

Yes No Yes

54 (Rossi et al., 2020b) 2020 Auto-scaling, resource Reinforcement Learning to calculate threshold

Solution

proposal

Yes No Yes

55 (Yudong et al., 2020) 2020 Resource Heuristics

Solution

proposal

No No Yes

56 (Zheng et al., 2019) 2019 Resource, auto-scaling Heuristics

Solution

proposal

No Yes Yes

57 (

Stefani

c et al., 2019) 2019 Resource, auto-scaling Architecture

Solution

proposal

No Yes Yes

58 (Herrera and Molt

o, 2020) 2020 Resource, auto-scaling Bio inspired algorithms

Solution

proposal

No No Yes

59 (Wang, 2019) 2019 Resource, auto-scaling Architecture

Opinion

paper

No Yes Yes

60 (Lakhan and Li, 2020) 2020 Resource, auto-scaling

Custom Algorithm; application partitioning

task ofﬂoading algorithm

Solution

proposal

No No Yes

61 (Rovnyagin et al., 2018) 2018 Container orchestration ML Engine for metrics analysis

Opinion

paper

Yes Yes Yes

62 (Sami et al., 2020) 2020 Resource, Context Architecture, custom algorithm

Solution

proposal

No Yes Yes

ICSOFT 2021 - 16th International Conference on Software Technologies

532