The collaboration tools offered by Cloud Computing have increased the need to share data and services within companies or between autonomous organizations. This has led to the deployment of community cloud infrastructures. However, several challenges will arise from this grouping of heterogeneous organizations. One of the main challenges is the management of trust between the actors of the community. Trust issues arise from the uncertainty about the quality of the resources and entities involved. The quality of a resource can be examined from a security or functional perspective. Therefore, ensuring security and monitoring the quality of resources is to ensure a high level of trust. Therefore, we propose in this paper a technique for dynamic trust management and quality monitoring of resources shared between organizations. Our approach consists, on the one hand, in evaluating the quality of resources based on quality of service measurement attributes and, on the other hand, in updating the trust values according to the information deduced from these measurements. The proposed framework is evaluated in terms of resource sharing success rate and execution time. Experimental results and comparison with TNA-SL and InterTrust models show that the framework can identify and track the behavior of malicious organizations with relatively low execution time.
According to the American consulting and research firm Gartner, 85% of companies will use Cloud Computing services by 2025 with 95% of their workload stored in the cloud compared to 30% in 2021 [
Several organizations with common needs and interests come together around community cloud infrastructure to minimize investment costs and promote resource sharing [
These trust issues in this type of environment can be seen from several aspects including the quality of shared resources. The Cloud Services Measurement Initiatives Consortium (CSMIC) has defined a set of qualitative and quantitative attributes for measuring the quality of Cloud Services (SMI) [
In environments such as the Community Cloud, the ability of organizations to guarantee and deliver quality resources over the long term conditions the sustainability of the infrastructure and its productivity. It is therefore essential to have mechanisms to track quality throughout the resource usage cycle. In this paper, we propose a framework for resource quality monitoring and dynamic trust management in a community cloud. We address the quality assessment of shared resources based on four (4) SMI attributes: response time, availability, threat and vulnerability management, and billing compliance. The main contributions of this paper are:
-A framework for identifying, defining and monitoring resource quality;
-A dynamic trust management mechanism derived from the results of resource sharing and usage in accordance with the established SLA.
The remainder of this article is organized as follows. Section 2 reviews related work. Section 3 describes the proposed framework. Section 4 presents the experiments and associated results. Finally, Section 5 concludes the paper and proposes perspectives for future work.
Managing resource quality and trust within distributed architectures remains a major concern for enterprises. Several works presenting techniques for the evaluation and comparison of services and their owners have been carried out.
K. Papadakis et al., proposed in [
In [
A secure resource allocation system (MSMC) among multiple organizations within a community cloud is proposed in [
S. Garg et al. presented SMICloud in [
In [
The literature studied shows that most of the work highlights the impact of service level agreements in the choice of resources or service providers. Furthermore, these works were carried out in federated or public cloud deployment environments that do not take into account certain requirements and governance modes of a community cloud. In addition, these proposals do not highlight the monitoring of trust and quality during the entire usage cycle of a resource. Therefore, it is necessary to propose a framework for trust monitoring and quality assessment of resources based on measurement attributes that are consistent with the social and sustainability characteristics of a community cloud environment.
The Community Cloud aims to enable organizations to share resources in order to reduce capital costs, create business opportunities without disregarding the quality of shared resources. We propose a framework for resource quality monitoring and dynamic trust management between organizations in the community. Resource quality monitoring is done through performance indicators offered by the CSMIC SMI attributes. A community cloud is characterized by organizations with specific needs or common interests. Thus, the active participation of members in the life of the community and the existence of lasting relationships are assets for the infrastructure. Furthermore, the social character of the collaborations between organizations is one of the fundamental reasons for a community cloud. Based on these facts, our framework is based on four (4) SMI attributes: availability, vulnerability level, response time and billing mode. The attributes of availability, vulnerability, and response time measure and ensure active participation of members and security of resources. In addition, the attribute of the billing mode of the resources brings into play the social aspect in this case the gratuity in the exchanges.
· Expression of needs and definition of SLAs: This layer houses the directory of resources and associated suppliers. In addition, it manages the service level agreements between the resource requester and the provider. The agreements are established on the basis of the four (4) SMI attributes mentioned above. In addition, it hosts the trust value manager and the inventory management module responsible for updating the availability status of resources.
· SLA monitoring: This consists of monitoring the state of the resources throughout the period of use in order to verify the conformity of the SLAs with the initial agreements established. This operation makes it possible to collect and consolidate the qualitative and quantitative contractual parameters related to the shared resource. This ensures that the commitments of each entity are respected.
· Update Manager: The role of the update manager is to update the trust information based on the information provided after the SLA monitoring. Each time the contractual parameters are violated during the period of use of the resource, the trust value update mechanism is triggered. In addition, SLA monitoring provides updated data for resource inventory management.
The CSMIC recommends a clear and simple definition for each attribute. Thus it proposes definitions centered around the following fields: measure name, related attribute, context, purpose, measure audience, measure definition, and data collection [
| Name of the measure | Billing method |
|---|---|
| Related attribute | Billing process |
| Context | When negotiating for resource sharing, it is important to It is necessary to specify the billing method (free of charge, sharing in exchange for a resource, exchange by means traditional fiduciary, exchange by virtual currency). The billing mode will have an impact on the willingness to share or not, on the quantity shared, on the time given to the use of the resource and on the extension capacities of the resource. |
| Measure audience | Organizations (supplier and applicant) |
| objective | It is a mutual consent measure of both parties. involved and which will allow or not the feasibility of a transaction. The commercial objective is the selection of a furnace provider that meets the financial needs of an applicant and vice versa. |
| Definition of the measure | Assign a value to each billing method. The values range from 1 to 4: 1) for the fiduciary mode-. 2) for a virtual currency, 3) for an electronic currency exchange for a resource and 4) free. Resources may be acquired free of charge depending on the reputation of the of the applicant or in exchange for a resource in order to encourage sharing within the community. |
| Data collection | The billing method is collected from each actor. |
| Name of the measure | Response time |
|---|---|
| Related attribute | Service Response Time |
| Context | A date on which the requester would like to have the response is associated with each request for a resource. The response time represents the time taken by the provider to respond to this request. |
| Measure audience | Organizations (supplier and applicant) |
| objective | Ensure the efficiency and timeliness of the implementation of the provision of a resource by a supplier. |
| Definition of the measure | The evaluation criteria are as follows: the date desired by the applicant tui, the response date promised by the tpj and the actual delivery date td. The time of response will be evaluated by three different scores: · The score (1) for an exchange made on the scheduled date; · The score (0.5) for an exchange that is scheduled before the desired date but is provided at a date greater than the scheduled period and less than the desired date; · The score (−1) for a delivery beyond the promised date and with a desired date less than the promised date. |
| Data collection | The dates (desired and promised) are defined respectively by the applicant and the supplier. The delivery date is deducted at the time of delivery of the resource. |
| Name of the measure | Availability |
|---|---|
| Related attribute | Availability |
| Context | A resource is provided for a defined period. It must have the capacity to maintain an agreed level of availability among the actors during the period of use of the resource |
| Measure audience | Organizations (supplier and applicant) |
| objective | Maintain an efficient level of resource availability provided within the community. |
| Definition of the measure | It is expressed as a proportion of time during which resources are available in relation to time during which they should be. Availablity ( r ) = T w T f (1) with Tw the service availability time and Tf the time total use of the service. This value will be compared to the promised availability rates by the supplier to assess the quality of availability of the resource |
| Data collection | The availability rate is obtained thanks to the returns of information from both actors |
| Name of the measure | Threat and vulnerability management |
|---|---|
| Related attribute | Proactive threat and vulnerability management |
| Context | The level of vulnerability of shared resources has an impact on the security of the infrastructure. To protect against threats and attacks, it is important to reduce the level of vulnerability of shared resources |
| Measure audience | Organizations (supplier and applicant) |
| objective | The goal is to ensure that shared resources meet the vulnerability levels of the CVSS [ |
| Definition of the measure | Assign sensitivity levels to each shared resource. These levels are defined according to the CVSS v2.0 vulnerability level standard |
| Data collection | The level of sensitivity is defined by each supplier of resources and validated by the applicant. |
The inventory management module updates the availability status of resources. The resource directory is updated when a new member joins, a new service is added, an organization leaves or a resource is no longer in use. An available resource is in an I and B status where applicable.
As in [
γ r ( t ) = { 1 if the resource is availlable then in a state I 0 else then in a state B (2)
The total quantity of a resource of vulnerability level j available at time t is expressed as follows:
q a j ( t ) = ∑ n = 1 Q r p j r s j ( t ) with r s j ( t ) = 1 if γ r n ( t ) = 1 else r s j ( t ) = 0 (3)
With Qrpj the total quantity of a resource of a vulnerability level j, and rsj(t) the availability state of the resource at time t. At the beginning of each new exchange, an update operation of the resource availability state is performed.
The selection of a resource provider for an exchange is based on the degree of trust the requester places in the provider. This trust value is calculated based on previous interactions between the organizations and the reputation of the provider. This value is expressed as follows:
g t = μ r d t + ( 1 − μ ) r e p (4)
With rep the reputation of the supplier (the initial reputation of an organization repini = 0), r d t the trust value resulting from previous interactions between the involved entities. This trust value is computed using the subjective logic (SL) presented in [
{ b = s s s s + s f + 2 d = s f s s + s f + 2 u = 2 s s + s s + 2 ⇔ { s s = 2 b u s f = 2 d u (5)
ss the number of successful prior shares between Or and Op, and sf the number of negative exchanges. The base rate α is defined by:
α = 0.5 (6)
The confidence value between Or and Op is expressed as:
r d t = b + ( α ∗ u ) (7)
For managing exchanges with an intermediate organization Oi between Or and Op, The SL proposes an update operator (⊗) to determine the transitive trust between organizations Oi and Op [
ω O p O r : O i = ω O i O r ⊗ ω O p O i { b O p O r : O i = b O i O r b O p O i d O p O r : O i = b O i O r d O p O i u O p O r : O i = d O i O r + u O i O r + b O i O r u O p O i α O p O r : O i = α O p O i (8)
Moreover, in the case of an exchange involving several two intermediaries Oi1 and Oi2 the trust between Or and Op, is deduced thanks to the consensus operator ( ⊕ ) and formulated as follows [
ω O p O i 1 : O i 2 = ω O p O i 1 ⊕ ω O p O i 2 { b O p O i 1 : O i 2 = b O p j O t z 1 u O p j O t z 2 + b O p j O t z 2 u O p j O t z 1 / d e n o d O p O i 1 : O i 2 = d O p O i 1 u O p O i 2 + d O p O i 2 u O p O i 1 / d e n o u O p O i 1 : O i 2 = u O p O i 1 u O p O i 2 / d e n o α O p O i 1 : O i 2 = α O p O i 1 (9)
d e n o = u O p O i 1 + u O p O i 2 − u O p O i 1 u O p O i 2
After the delivery of the requested resource, an initial evaluation of the exchange is performed. This evaluation allows an initial update of the trust values and reputation. In addition, during the use of the resource and until the end of the period of use, this process of updating the values is triggered by each violation of the quality of service attributes specified in the SLA.
As in proposal [
r e p O p = { r e p O p c t + θ n if positive result r e p O p c t + θ n / 2 if positive result with minor violation r e p O p c t − θ n if negative result with θ n ( k ) = ( V lmin ( k ) + V lmax ( k ) ) / 2 V lmax (10)
r e p O p c t the current reputation of the provider, Vlmin(k) the minimum value of vulnerability degree of the resource for a level k, Vlmax(k) the maximum value of vulnerability degree, Vlmax the maximum value of vulnerability degree. The resource vulnerability levels are defined according to the CVSS 2 standard [
k = 1 if V l ∈ [ 7 , 10 ] k = 2 if V l ∈ [ 4 , 7 [ k = 3 if I l ∈ [ 0 , 4 [ (11)
The performance of our proposal was evaluated through simulations of resource sharing between organizations in a community cloud. Data was generated to model the organizations and resources in the community. A MacBook Pro computer (Retina, 15-inch, mid-2015), 2.2 GHz Intel Core i7 quad-core processor,
16 GB 1600 MHz DDR3 memory was used to perform our experiments. The python language was used for programming with a Pycharm editor. Organizations providing SLA compliant resources are qualified as good providers. Those providing non-compliant or violating resources are called malicious. To examine the scalability of our framework the simulations are conducted on organization groups of 80,120, 180,220 and 250 members. Furthermore to measure the attack resistance of our model, the number of malicious providers is varied from 20%, 40%, 60%, 80%. The performance of our model is compared with two other models namely TNA-SL [
SSR = number of resources provides by good organizations total number of resources provided by good organizations (12)
In addition, the dynamic management of trust through the updating of trust values based on the respect of SLAs, allows to increase the reputation values of the organizations contrary to the two other models which do not integrate this aspect. The SSR of the InterTrust and TNA-SL models also experience a relatively lower evolution for rates of 20%, 40% and 60% of malicious illustrated by
The community cloud deployment model promotes collaboration and resource sharing (data and services) between organizations with specific requirements and common needs. However, ensuring trust between members of this community
remains a major challenge. Monitoring the quality of shared resources in this environment is an answer to this problem. Therefore, we propose in this paper a framework for resource quality evaluation and trust monitoring based on quality of service measurement attributes. The model is based on the definition of a service level agreement based on four SMI (Service Measurement Index) attributes: billing mode, availability, threat and vulnerability management and response time. Moreover, the monitoring of SLA attributes allows to propose a dynamic management mechanism and to update the trust and reputation values. Through experiments, our proposal has been compared to the TNA-SL and InterTrust models. The results of the experiments show that our proposal allows the deployment of a community cloud that is resistant to attacks. On the other hand, it allows distinguishing good providers from malicious ones and to eliminate or reduce the participation of malicious ones in the community exchanges. Finally, our model offers better execution times compared to TNA-SL and Intertrust models. In future work, we will propose an agent-based collaboration framework. The aim is to present a model characterized by explicit communication and mutual commitment of the actors for successful exchanges.
The authors declare no conflicts of interest regarding the publication of this paper.
N’goran, R., Vallee, L.N., Johnson, G.Y. E., Tetchueng, J.-L., Kermarrec, Y. and Asseu, O. (2022) Shared Resource Quality Monitoring and Dynamic Trust Management in a Community Cloud. Open Journal of Applied Sciences, 12, 1898- 1914. https://doi.org/10.4236/ojapps.2022.1211131