EE-(m,k)-Firm: A Method to Dynamic Service Level Management in

Enterprise Environment

Bechir Alaya

1,2

IResCoMath-Lab, University of Gabes, FSG, Erriadh street 6072 Zrig Gabes, Tunisia

Higher Institute of Technological Studies, University of Gabes, Tunisia

Keywords: Enterprise Management, Operations Analytics, QoS, (m,k)-Firm Model.

Abstract: Due to enterprises environment specificities, the operations management is actually a challenging problem.

In this paper, we choose to using a compromise between the available resources and the quality of service

(QoS) granularity. We join this compromise to a guaranteed technique in order to reach an intelligent loss of

Sub-operations according to the importance of each operation. The resulting approach permits the increase of

availability, performance, reliability and system dependability. The aim of our contribution is to ensure the

client satisfaction by increasing the QoS while dealing with the enterprises environment characteristics. The

effectiveness of what we propose is measured through a simulation study.

1 INTRODUCTION

In the last few years, Enterprise Performance

Management (EPM) is became an integrated business

intelligence solution that gives companies a

comprehensive view of their enterprise. Many of the

challenges that companies face today, which greatly

influence the enterprise performance and its quality

of service (QoS).

These challenges are specified by completely

uncontrollable variables, that show the current

operation of the system and its performances

compared to the optimal performance required by

managers. These variables considerably show the

current operation of the system and its performances.

Enterprise modeling remains always a challenge,

despite the significant advances in modeling

technology. The modeling for different points of the

company is necessary. Such a modeling is part of the

answer to the need for integrating the production

functions and specially the maintenance and QoS

guarantee. The policy that we propose can be

generalized and therefore applied to all the enterprise

functions.

Due to the similarities between data management

in RTDBS (Hamdaoui and Ramanathan 1995)

(Decker, 2014) and in enterprises (Barnes et al.,

2015), we propose to adapt some results obtained on

the management of QoS in the RTDBS to manage the

performance of companies. However, we present a

model based on (m, k) -firm model (Davide et al.,

2012) (Wang et al., 2004) (Cho et al., 2010)

(Goossens, 2008) studded in RTDBS to take into

account the congestion of systems workload in firms.

Our main objectives were to design a model that

meets the performance requirements of customers

and managers and provide QoS guarantees and

robustness when customer requests grow rapidly or

when company resources are congested.

This paper is organized as follows. Section 2

describes some work related to the management of

service quality; Section 3 shows the (m, k) -frim

model; Section 4 explains the different characteristics

of EE- (m, k) -firm model; Section 5 describes a

method for automating the processing of EE- (m, k) -

firm constraints; Section 7 analyzes the simulations

results and finely, Section 8 presents the Conclusion

and some remarks.

2 RELATED WORK

The QoS management in enterprise environment (Li

et al., 2006) (Partha et al., 2014) (Arboleda et al.,

2016) is a typical problem. Recently, several studies

have been based on this topic. In (Arboleda et al.,

2015), the authors proposed that to reach a superior

performance, it is necessary to suggest (i) the

114

Alaya, B.

EE-(m,k)-Firm: A Method to Dynamic Service Level Management in Enterprise Environment.

DOI: 10.5220/0006322401140122

In Proceedings of the 19th International Conference on Enterprise Information Systems (ICEIS 2017) - Volume 1, pages 114-122

ISBN: 978-989-758-247-9

adoption of appropriate strategic behaviors to client,

to competitor and to technology and (ii) the targeting

of the appropriate market segments, notably early

adopters, innovators, early majority, laggards and late

majority. In their proposition, the strategic behavior

of corporate performance relationship is subject to the

company's strategy by examining this relationship on

high technology markets and considering further

contribution of the appropriate target market

selection. This approach provides useful orientation

to business managers to the steps that they should take

to augment their performances.

The authors of (Barnes et al., 2015) accentuated the

effect of interpersonal factors on company's

performance through the relationship quality and the

intervening roles of intercompany trust. The authors

justified that trust plays an instrumental role in

enhancing the components of the inter-firm

relationship quality. They showed that inter firm

relationship quality is positively related to superior

financial performance, and most of the associations

between each of the interpersonal factors and inter

firm trust were moderated by the importer's size and

foreign supplier's origin as well as the length of the

relationship and which party initiated the relationship.

In (Slater et al., 2007), the authors proposed a

technique to improve the risk analysis in Enterprise

Resource Planning (ERP) (Li et al., 2006) (Mehrjerdi,

2010). They aimed to obtain a more structured

systematic model of the different relationships

between the risk factors/effects associated with ERP

projects and attain a better understanding. The major

objectives of their work were to (i) allow a

collaborative approach to risk analysis, (ii) help the

administrators in treating and controlling project risk

and (iii) help the administrators to comprise the links

between the development of a relevant risk analysis

strategy and the evaluation of a global risk index for

each factor used.

Demand Response Management (DRM) which is a

key component of the future smart gridDemand

Response Management (DRM) was the subject of

(Chai and Chen, 2014). In fact, the author studies

DRM with different public service companies.

Depending on the requirement of enterprise,

several types of information systems have been

improved for various goals (Krell et al., 2016). A

study in (Al-Mamary et al., 2014) attempted to

demonstrate the role of each type of information

systems in firms' organizations. According to

O’Brien & Marakas (Brien and Marakas, 2010), the

applications of information systems that are

implemented in today’s business world can be

classified in several different ways. In enterprises

world, there are varieties of information systems such

as, Office Automation Systems (OAS), Expert

System (ES), Transaction Processing Systems (TPS),

Management Information Systems (MIS), Executive

Information Systems (EIS), Decision Support System

(DSS), etc. Each type of information system has a

specific objective in management operations and in

organizational hierarchy (Alam et al., 2015).

Other researches were proposed in (Kadiri et al.,

2016) (Atkinson et al., 2015) to present customized

views of enterprise systems to various stakeholders

according to their competencies and requirements.

For a better QoS, they were interested in developing

and improving the services and languages offered by

such tools on a continuous basis. They discuss the

weaknesses and strengths of different approaches

(Nikolow et al., 2013) interested in language

development and proposed a modeling framework

more able to support the main extension scenarios

currently found in practice.

3 ENTERPRISE AND QoS

MANAGEMENT

3.1 (m,k)-Firm Model

The recurrence of tasks in real-time systems allow to

ignore some invocations (or jobs) using (m,k)-firm

constraints. These constraints specify that in a

window of k invocations, at least m tasks (0 ≤ m ≤ k)

must respect these deadlines (West and Zhang, 2004)

(Hamdaoui and Ramanathan 1995) (Cho et al., 2010).

Otherwise, for k tasks, m tasks are required and (k-m)

tasks are optional. In (Bernat, 1998), Bernat showed

through an example why it is best to use two

parameters to define this type of constraints.

Furthermore, it has been shown that the concept of

(m,k)-firm constraints is appropriate for specification

(management) of QoS of real-time application (Wang

et al., 2002). To effectively manage the tasks under

(m,k)-firm constraints, new scheduling algorithms

have been proposed (Hamdaoui and Ramanathan

1995) (Ramanathan, 1999). They are divided into two

main groups units, (1) Dynamic algorithms

(Hamdaoui and Ramanathan 1995) and (2) Static

algorithms (Dixon and Verma, 2013). Briefly, the

static algorithms provide a deterministic vision of the

system, while the dynamic algorithms rather provide

a probabilistic vision. The dynamic algorithms take

into account any system modification.

EE-(m,k)-Firm: A Method to Dynamic Service Level Management in Enterprise Environment

115

3.2 (m,k)–Firm Constraints

Application

The enterprises management aims to meet the client

requests by integrating many constraints: the costs,

products quality, deadlines, customer demand,

necessary personnel, infrastructure, supply of raw

materials, etc.

The enterprise management must take into account

four main types of constraints (Dixon and Verma,

2013).

The main objective in enterprise environment is

both quantitative and qualitative. The QoS

degradation implies the degradation of system

performance in such a way that the system continues

to function but with a disequilibrated level of QoS. In

an overload situation, the production and QoS

degradation is inevitable since clients' demands will

always be dropped or delayed although many clients'

demands can tolerate some delay if they arrive with a

permitted mode. Moreover, the effect on QoS in

enterprise environment depends on how and when the

degradation is present.

The proposed method can be described as follows:

a task in process of industrial production is

constrained by (m,k)-firm requirements if at least m

task instances within a range of k consecutive tasks

respect their intended deadlines. If more than (k−m)

deadline of tasks fail in k consecutive tasks at that

moment, we can mention that the tasks will fall in a

dynamic failure state. Consequently, the QoS

constraints will not be satisfactory for the customers.

For each enterprise branch, the values of m and k vary

according to the criticality of tasks and system load.

In practice, the values indicated by the industrial

systems are not all of the same importance.

3.3 Tasks Management and Adaptation

The tasks of an enterprise system are decomposed

into several classes according to their tolerance to

tasks loss.

We consider three classes of tasks in the industrial

environment: critical task, hard non-critical task and

optional task. With this technique, which we called

(m,k)-firm in Enterprise Environment (EE-(m,k)-

firm), we can realize a compromise between the

available resources and the QoS granularity in the

same type of task.

In this work, we focused on the adaptation of the

number of tasks to the system load state. We assumed

that measures of the system capacity were available

on the one hand and that we had a significant number

of client demands on the other hand.

We also assumed a system situation in a production

enterprise, whose actual performance is N, was

overloaded. We supposed that Optimal QoS (Opt-

QoS) was the quality of the client demand

necessitating M tasks. In order to be coherent with the

system performance, it was requisite to throw (M−N)

tasks. Consequently, we had to reduce the quality of

the client demand and if necessary, we could remove

some tasks. However, the removal without applying

a control method would be arbitrary.

The removed tasks are lost from the system,

causing QoS degradation, notably if some critical

tasks (Goossens, 2008) are removed. In this work, we

adapted the EE-(m,k)-firm constraints, that serve to

discard some tasks but intelligently.

The three classes of tasks were proposed to adjust

the QoS requested by the clients based on real system

capacity. We proposed that constraints for each task

category were fixed as follows: EE-(mc,kc)-firm for

critical tasks, EE-(mh,kh)-firm for hard non-critical

tasks and EE-(mo,ko)-firm for optional tasks.

Notably, mc tasks must be executed among kc tasks.

The system capacity was calculated using the

formula: mc + mh + mo, where mc and kc present the

constraints of critical tasks .The constraints of

different task classes are organized as follows:

mc>mh>mo. In the enterprise environment, we

usually propose that mc=kc, given that these types of

tasks are critical and that it is not recommended to

lose them.

With the application of our EE-(m,k)-firm policy,

we suppose that the required capacity necessary to

respond to an enterprise transaction is M. With:

M=kc + kh + ko,

N=mc + mh + mo.

We proposed how to equilibrate the QoS at the

tasks level in a production enterprise according to the

available system capacity. We began by calculating

the required capacity by all the current clients. Then,

we calculated the rate that presents the ratio between

the available system capacity (N) and the required

capacity.



Rate

)

Given that:

 k present the number of available tasks in the

system.

 DRi present the demanded resource by task i.

ICEIS 2017 - 19th International Conference on Enterprise Information Systems

116

4 INTELLIGENT TASKS LOSS

In order to guarantee an intelligent loss of tasks

according to their importance, we defined a method

which describes how a client demand is composed of

k tasks. These tasks follow a well-defined

organization when the system resources are not

available to respond to all necessary tasks to have the

requested QoS.

Given the differentiation between tasks, we

already proposed that the critical tasks are two

categories. In one part, the mandatory and critical

tasks (C1) without which the client demand will not

be realizable. In the other part, the critical tasks (C2)

without which, the client demand can be realizable,

but the QoS will be extremely poor.

Similarly to the hard and non-critical tasks, we

proposed a classification into two categories, namely

H1 and H2. Finally, for optional tasks that usually

reflect the QoS degree we have two classes (O1, O2)

according their importance for QoS level.

Consequently, The k tasks of a client demand

based on EE-(m,k)-firm constraints are represented

by a succession of k elements from

{C1,C2,H1,H2,O1,O2} as described previously.

In enterprise environment, it is difficult to have an

approach that guarantees an optimal QoS for all

clients arriving to the system. Using this intelligent

loss specification, each client can show these EE-

(m,k)-firm constraint according to requested QoS and

the available system resources. A minimum QoS is

guaranteed if at least all critical and mandatory tasks

are executed. Notably, if some optional tasks are

missed by the system, the degradation will be only to

EE-(mo, ko)-firm constraints, but not to EE-(mc, kc)-

firm constraints.

These constraints are extremely appropriate in

order to extract all requirement of a client demand. In

all cases, a client demand is represented by a

succession of critical tasks, hard tasks and optional

tasks.

The loss of tasks in a type of critical tasks and/or

in a type of hard tasks necessary for a client demand

will cause some degradation in the following tasks

until a new demand occurs, although the optional

tasks loss has no effect.

4.1 Dynamicity of EE-(m,k)–Firm

Constraints

To improve performance, availability, reliability, and

system dependability, we applied a method of

dynamic treatment of tasks. The objective was to

automate the treatment of EE-(m,k)-firm constraints.

The need of responding to critical and hard tasks

is most crucial when it comes to sensitive systems

where an error can be humanly costly other than

financially. This is the case, for example, for nuclear

reactors, chemical factory, aircraft systems and many

others. Following to importance of client demand and

the importance of meeting their QoS requirements,

we associated a dynamic analysis modules to EE-

(m,k)-firm constraints, to optimize the gap between

the rate of received QoS and the rate of desired QoS.

In some situations, the system stops the operation of

any other tasks (critical, hard or optional) to respond

to tasks that are humanly more critical.

A deviation detecting module between the

provided QoS by the system and the desired QoS by

the client has become necessary. Then, according to

task type, a consultation of gap impact must be

carried out. Finally, the system decides the necessary

values of EE-(m,k)-firm constraints.

4.2 Detection and Localization Gap

The detection procedure aims to determine the

occurrence of a gap between the values m and k of

each task type that has a specific EE-(m,k)-firm

constraints fixed by the system. Indeed, because the

properties of different tasks according to their types,

the difference between MC and KC is more important

than the difference between MH and KH and also

between MO and KO. However, this detection

procedure will be applied to all possible types of

constraints. Generally, for proper operation of an

enterprise, these differences are usually of zero mean,

which represents an optimum QoS to clients.

A means to auto-observe the gap between

different EE-(m,k)-firm constraints is to estimate the

needed values for each constraints type (KC, KH and

KO). The estimated values of MC, MH and MO are

then respectively subtracted from maximum

constraints KC, KH and KO to form the gaps E(C),

E(H) and E(O) as follows:











M-

K =E(O)

M-

K =E(H)

M-

K =E(C)

Given that KC> MC, KH> MH and KO> MO.

At production times in an enterprise, the gap E(·)

will significantly deviate according to the increase of

system load, it will be equal to zero except when the

system operates normally. In real applications, the

differences are not exactly a zero value for the

systems absence that accurately reflects the actual

state of resources. Besides, the assigned

measurements that aim to reflect the available

resources are often marked by measurement noises.

The optimal QoS for clients varies according to

values of measurement noises. With the proposed

treatment of EE- (m,k)-firm constraints (see figure 1),

EE-(m,k)-Firm: A Method to Dynamic Service Level Management in Enterprise Environment

117

Figure 1: Treatment structure.

and depending on the criticality of supplied products

(chemical, nuclear ...), the values of KC and KH must

be accurately measured. The gaps will then be written

as:











M-

K =E(O)

M-

K =E(H)

M-

K =E(C)

Where Km (·) is the value measured by the system

and which is composed by the real value K(·) and the

various types of noises relating to the calculation

uncertainties .

To guarantee the application of EE-(m,k)-firm

constraints, we propose a comparison method of each

gap E (.) at an optimal predefined threshold for each

type of task: Threshold ε for critical tasks, ε' for hard

tasks and ε'' for optional tasks, respectively. At every

crossing of threshold, an alert is sent to the system for

a new QoS management, we will then have:







↔

↔≤

1 =Alert > E(C)

0 =Alert E(C)







↔

↔≤

1 =Alert ' > E(H)

0 =Alert ' E(H)







↔

↔≤

1 =Alert '' > E(O)

0 =Alert '' E(O)

After detecting the presence of a gap between M

and K, it is necessary to locate the task type affected

by this gap. This is nominated by the gap localization.

At the realization, we proceed at a structuring of

all generated gaps during the system function.

Generally, we constructed a first set of gaps Ei (·) that

depend on the tasks types. From these basic gaps, we

form two types of gaps: hard gap and soft gap.

In case of hard gap, after receiving an alert, the

system immediately acts even by an intelligent

violation of allocated resources to other clients’

demands. This gives a dynamicity of resources

allocation and EE-(m,k)-firm constraints. However,

in case of soft gap, the system does not immediately

act, but waits for the availability of resources to

respond to this task type. During system function, the

EE-(m,k)-firm constraints dynamically vary

according to priority of client demands, system load

and gap type.

We will have a decrease in optional tasks number

and an increase in critical tasks number. For hard

tasks, the number varies depending on the decrease

and increase of critical and optional tasks.

5 SIMULATIONS AND RESULTS

We now detail the implementation of the EE-(m,k)-

firm policy. Four types of decisions should be taken

by our policy. We first describe the necessary data

structures, and then we consider each of these

decisions separately.

5.1 Description of Data Structures

Table.1 shows the data structure for each client

demand. In a table noted table of demands in which

each line contains the tasks number of a demand, and

the class of popularity (EE-(m,k)-firm constraints),

indeed, three classes are present.

The first refers to the C tasks (Critical) which are

the most requested tasks by the system. The second

regroups tasks of average importance H (Hard). The

third contains optional L tasks (Low), i.e. least

required by the system. Tasks table (table 2) records

various information about the demanded tasks. Note

that, the demands may not have the same number of

Table 1: Demands table.

Demand-id

Requested

tasks

EE-(m,k)-firm

constraints

C H L

Demand_1 14 4 4 6

Demand_2 7 3 2 2

Demand_3 9 4 3 2

Demand_4 4 3 1 0

tasks. Each entry in the units table (table 3)

corresponds to a unit and maintains several counters

that keep track of free and served resources.

ICEIS 2017 - 19th International Conference on Enterprise Information Systems

118

First, for load balancing on the units, the choice

of the tasks will be on a lightly loaded unit that is

selected for execution.

Table 2: Tasks table.

Number

of task on

execution

Number

of unit

tasks

Unit-id

Task_1

60 4

Task _2

70 2

Task _3

50 3

Task _4

20 4

This is achieved by traversing the entrance of the

tasks table to find all the units that contain the type of

requested task (including tasks in progress) and then

looking into the set of the corresponding units to find

the least loaded unit. Whereas unit Ui has completed

execution of a task Tj, the data structures must be

updated, to indicate a resource liberation on Ui. This

is done by resetting the counter in the entry of tasks

table.

Table 3: Units table.

Id-

Unit

Total

resource

(Rs)

Free

Served

Rs in free

admission

U1 1000 100 900 10

U2 3500 700 2800 5

U3 2000 300 1700 7

U4 1500 200 1300 20

5.2 Unit and Task-type Selection

After having taken the decision to get the process of

responding to a client demand, the EE-(m,k) -firm

policy must select the tasks using the different

constraints.

The execution begins when the system completes

the selection of different tasks types of a demand.

Note that the EE-(m,k)-firm policy does not change

the task type simply because a resource is released by

this task or because it has caused a unit overload.

This avoids the problem of changing the task type

which slightly affects the QoS requested by the client.

The estimated profit of Pti to execute a task i of a

demand is a measure of future load that can be

reduced from the current unit. This is calculated as

follow:



=



−



n









)

Where W represents the weighting factor. The

motivation for using this formula is to change the task

type where the advantage in terms of load is expected

to be higher in the future.

The load that can be changed in the near future

(execution task time) is given by the load on the

previous task. However, the load on the previous

tasks represents the load that can be changed

gradually in the future. To further improve the

performance of the immediate load transfer, the profit

to execute a task was calculated by weighting

exponentially.

Algorithm 1.

Ti = number of tasks of i type

Pti = Execution profit of task i

Si = unit number that can execute the

task i

Sth = threshold of tasks number

Pi = Popularity of i task

Pmin = min (Pi)

Pmax = max (Pi)

Pmoy = (Σi = 1..N Pi) / N

Class L = [Pmin (Pmin + Pmoy) / 2]

Class H =[(Pmin+Pmoy)/2,

(Pmax+Pmoy)/2]

Class C = [( max + Pmoy) / 2, max]

V (V1 ... ..Vj ...... .VN)

For (j = 1 to N)

If (Vi∈ H Class)

For (i = 2 to tasks number)

If (Si> Sth)

R = round ((Si-sth)/quota)

If (Pti superior to all benefit of

another tasks)then

Execute the task i in the first

selected unit

The load on the precedent tasks can be found from

entries matching tasks in the tasks table. Also, if there

are ti tasks of the current demand i, creating a

modification of task type result (1/(ri -1)/(ri + 1)) of

profit in terms of load movement. The current number

of tasks ri will be also available in the current task

entry.

5.3 Simulation Results

First, we discuss the impact of the tasks number on

the system response time (access delay) with ''fixed

EE-(m,k)-firm constraints'' and ''dynamic EE-(m,k)-

firm constraints''. Figure 1 describes the

representative results for different values of demands,

such as 10, 20, 30, 40 and 50.

Figure 2 shows that the response time for all

curves decreases with the increase of tasks number.

The time considerably decreases between a low tasks

EE-(m,k)-Firm: A Method to Dynamic Service Level Management in Enterprise Environment

119

number and an important tasks number. Indeed, the

load balancing between different tasks types is

significantly reduced. This is due to dynamics of tasks

treatment that affects several factors. In particular, the

access delay, in case of a unit overload improved with

a dynamic EE-(m,k)-firm constraints, since it

depends on tasks criticality that will be dynamically

treated by the system. Thus, the system, which will be

able to answer, has several tasks with these dynamic

treatments, leading to improve management of the

storage space and the QoS. This means that the EE-

(m,k)-firm policy brings several benefits other than

reducing the access delay.

Figure 2: Access delay.

At the tasks execution, the system begins the tasks

dispersion between the necessary units. The second

application of EE-(m,k)-firm policy results in the

correct application of dynamicity of the policy

showing that mC and kC have the highest priority.

The graph in figure 2 shows the behavior we

expected. We can equally notice from the same figure

that EE-(m,k)-firm policy, with fixed or dynamic

constraints in all loads requirement, gives a shorter

response time.

But, we note that when increasing the tasks number,

the difference between results decreases.

Consequently, we can predict that if the number of

tasks attains a certain threshold, there will be no

difference between the different algorithms.

Figure 3: Reject rate.

Figure 3 shows that our policy with a dynamic

treatment of EE-(m,k)-firm constraints significantly

reduces the rejection rate. The difference between the

curves, using fixed and dynamic constraints, shows

the improvement of tasks acceptance rate. The gap

between sub-curves of EE-(m,k)-firm policy with a

dynamic constraints m and k on different numbers of

demands, shows the effectiveness of this dynamicity

on the rejection rate. The ratio between the decrease

of rejection rate with the increase of demands number

shows that when the tasks number increases, the

curves of our policy will be confused. We can

conclude from these comparisons that our proposed

policy achieved the desired results, even with a large

tasks number.

The served tasks rate present the ratio among the

number of received and executed tasks and all

requested tasks.

We describe the case of little workload arriving to

the case of the best workload. With dynamic

constraints of EE-(m,k)-firm policy and in case of

weighty system workload, our policy substantially

affects the rate of served tasks. Consequently, at

different workloads, EE-(m,k)-firm policy is

powerful to overcome the system congestion

problems (figure 4). From this study, we notice that

the EE-(m,k)-firm policy provides satisfactory

results.

Outstandingly, EE-(m,k)-firm policy leads to an

important number of served tasks in the case of high

workload, up to 98% with dynamic constraints and

about 49% with fixed constraints.

ICEIS 2017 - 19th International Conference on Enterprise Information Systems

120

Figure 4: Served tasks rate.

6 CONCLUSIONS

The main purpose of this study was to present a novel

policy of a specific treatment technique of tasks in

enterprises environment. We proposed the EE-(m,k)-

firm policy to indicate the necessary tasks and

calculate their ranking, using a compromise between

the available resources and the QoS granularity in the

same task type. Based on an in-depth review of the

relevant literature, three categories of tasks are

possible in enterprise environment, namely critical

tasks, hard tasks and optional tasks.

Afterwards, a guaranteed technique was applied

to losses tasks intelligently according to importance

of each task. A dynamicity of EE-(m,k)-firm

constraints is then used to attain an increase of

availability, performance, reliability and system

dependability.

The results obtained from the proposed policy

reveal that a ‘‘lack of awareness regarding the

benefits of dynamic treatment of tasks in an

intelligent and dynamic manner in enterprise

environment is the most important reason behind the

implementation of EE-(m,k)-firm policy. This type of

policy can be extremely valuable for companies that

wish to focus their efforts and resources to guarantee

a satisfactory QoS for end-users and challenges

toward the successful implementation of tasks

management.

REFERENCES

Hamdaoui, M., Ramanathan, A., 1995. ''A Dynamic

Priority Assignement Technique for Streams with (m,k)-

firm Deadlines'', IEEE Trans. Computers, vol: 44, No:

12, pp: 1443-1451.

Decker, A., 2014. ''Factors Influencing The Quality of

Accounting Information System And Its Implications on

The Quality of Accounting Information'', Research

Journal of Finance and Accounting, Vol:5, No:2.

Barnes, B, R., Leonidou, L. C., Siu, N. Y.M, and Leonidou,

C. N., 2015. ''Interpersonal Factors as Drivers of

Quality and Performance in Western–Hong Kong

Interorganizational Business Relationships'', Journal of

International Marketing, Vol: 23, No: 1, pp: 23-49.

Davide, A., Riccardo, D., and Valeria, M., 2012. ''Risk

assessment in ERP projects'', Information.

Systems. Vol: 37, pp: 183-199.

Wang, Z. , Chen, J., Sun, Y., ''An integrated DBP for

streams with (m,k)-firm real-time garantee'', J.

Zhejiang Univ. Sci, Vol: 5, No: 7, 2004.

Cho, H., Chung, Y., Park, D., 2010. ''Guaranteed dynamic

priority assignment schemes for real-time tasks with

(m,k)-firm deadlines'', ETRI Journal, Vol: 32, No: 3,

pp: 422-429.

Goossens, J., 2008. ''(m k)-firm constraints and DBP

scheduling: impact of the initial k-sequence and exact

schedulability test'', In16th International Conference on

Real-Time and Network Systems. Rennes, pp: 61-65.

Li, J., Song, Y., Simonot-Lion F., 2006. ''Providing real-

time applications with graceful degradation of QoS and

fault tolerance according to (m,k)-firm model'', IEEE

Transactions on Industrial Informatics, Vol: 2, No: 2,

pp: 112-119.

Partha, P., Atighetchi, M. , Soule, N., Ishakian, V., Loyall

J., Grant, R., Sinclair, A., 2014. "Secure and QoS-

Managed Information Exchange Between Enterprise

and Constrained Environments",17th IEEE

International Symposium on Object /Component/

Service-Oriented Real-Time Distributed Computing

(ISORC), pp: 141-149.

Arboleda, H., Paz A., Jiménez, Tamura G., 2016.

''Development and Instrumentation of a Framework for

the Generation and Management of Self-Adaptive

Enterprise Applications'', Ingeniería y

Universidad, Vol: 20, num: 2, pp: 303-334.

Arboleda, H., Paz, A., Jiménez, M. and Tamura, G., 2015.

''A framework for the generation and management of

self-adaptive enterprise applications'', 10th Computing

Colombian Conference (10CCC).

Slater F., Tomas G., Hult, M., Eric M., 2007. ''On the

importance of matching strategic behavior and target

market selection to business strategy in high-tech

markets Stanley'', Journal Academy of Marketing

Science. Vol: 35, pp: 5-17.

Li, J., Song, Y., Simonot-Lion F., 2006. ''Providing real-

time applications with graceful degradation of QoS and

fault tolerance according to (m,k)-firm model'', IEEE

Transactions on Industrial Informatics, Vol: 2, No: 2,

pp: 112-119.

Mehrjerdi, Y. Z., 2010. "Enterprise resource planning: risk

and benefit analysis", Business Strategy Series, Vol:

11, No: 5, pp: 308 – 324.

Chai, B., Chen, J., Y. Zhang, 2014. ''Interpersonal Factors

as Drivers of Quality and Performance in Western–

Hong Kong Interorganizational Business

EE-(m,k)-Firm: A Method to Dynamic Service Level Management in Enterprise Environment

121

Relationships'', Smart Grid, IEEE Transactions

on, Vol:5, No: 2, pp: 722 - 731.

Krell, K., Matook, S., Rohde, F., 2016. ''The impact of

legitimacy-based motives on IS adoption success: An

institutional theory perspective'', Information &

Management, Vol: 53, No: 6, pp: 683–697.

Al-Mamary, Y. H. S., Shamsuddin A., and Aziati, A. H. N.

2014. ''The Role of Different Types of Information

Systems In Business Organizations: A Review'',

International Journal of Research (IJR), Vol: 1, No: 7.

Brien, J.A., Marakas, G.M., 2010. ''Management

information systems'', 10th ed., byMcGraw-Hill/Irwin,

a business unit of The McGraw-Hill Companies.

Alam, K. A., Ahmad, R., Akhunzada, A., Nasir, M. H. N,

Khan, S. U., 2015. ''Impact analysis and change

propagation in service-oriented enterprises: A

systematic review'' Information Systems journal, Vol:

54, pp: 43-73.

Kadiri, S. E., Grabot, B., Thoben, K., Hribernik, K.,

Emmanouilidis, C., Cieminski, G. V., Kiritsis, D.,

2016. "Current trends on ICT technologies for

enterprise information systems", Computers in

Industry, Vol: 79, pp: 14-33.

Atkinson, C., Gerbig, R., Fritzsche, M., 2015. ''A multi-

level approach to modeling language extension in the

Enterprise Systems Domain'', Information.

Systems. Vol: 54 , pp: 289-307.

Nikolow, D., Slota, R., Polak, S., Mitera, D., Pogoda, M.,

Winiarczyk P., Kitowski, J., 2013 ''Model of QoS

Management in a Distributed Data Sharing and

Archiving System'', Procedia Computer

Science, International Conference on Computational

Science 'ICCS', Elsevier, pp: 100-109.

West, R., Zhang Y., 2004. Dynamic window-constrained

scheduling of real-time streams in media servers. IEEE

Trans. Comput, Vol: 53, No: 6, pp:744-759.

Bernat, G., 1998. ''Specification and Analysis of Weakly

Hard Real-Time Systems'', PhD thesis, Université de les

Illes Balears.

Wang, Z., Song, Y., Poggi, E.M., Sun Y., 2002. ''Survey of

Weakly-Hard Real-Time Schedule Theory and its

Application'', Distributed Computing and Applications

to Business, Engineering and Science (DCABES), pp:

429-437.

Ramanathan,, P., 1999. ''Overload Management in Real-

Time Control Applications Using (m,k)- Firm

Guarantee'', IEEE Trans. Parallel Distrib. Syst, vol: 10,

No: 6, pp: 549-559.

Dixon, M., Verma, R., 2013. ''Sequence effects in service

bundles: Implications for service design and

scheduling'', Journal of Operations Management, Vol:

31, No: 3, pp: 138-152.

ICEIS 2017 - 19th International Conference on Enterprise Information Systems

122