A Semantic Framework
for Cloud Resource Management in Business Processes

We develop a framework that provides a semantic description of cloud resources and strategies to resolve conflicts in business process development. This framework aims at ensuring a correct and controlled use of resource allocation with a free-of-conflict. To this end, the framework supplies a common understanding which corresponds to a semantic and unified description of cloud resources, policies to assign cloud resources to business processes based on this description, and a set of social relations connecting business processes components namely task, person, machine, and resource. A general explanation of this framework is given through an exploratory paper.
Basically, Our contributions include:

  1. A framework that provides a semantic description of cloud resources
    1. Strategies to assign cloud resources to business processes based on this description
    2. A set of social relations connecting BP's components including resources
    3. Strategies to resolve resource conflicts.
  2. A proof of concept demonstrating the feasibility of this framework.

Cloud Resource Description:
The cloud offers three main types of resources at the Infrastructure-as-a-Service (IaaS) model which can be spread into virtual machines (VMs). They consist of compute, storage, and network resources.

Compute Resources
The Compute resource are a collection of Physical Machines (PMs) where each contains one or more processors, memory, network interface and local I/O (e.g., virtual machines). They represent information processing resources.
Storage Resources
The Storage resources depict persistent storage services where each service have varying levels of data consistency and reliability.
Network Resources
The network resources serve to interconnect physical machines with a high bandwidth network.

Framework Overview:
Our framework for conflict resolution of resources considers (i) the structured nature of the business world in terms of business processes and (ii) the unstructured nature of the social world in terms of messages shared like tweets, spontaneous relations established between users, etc. Our framework taps into the social data that could help deploy business processes over the cloud. By social data we mean social relations that exist between tasks, between resources, between cloud providers, between users, and between all of them. For instance, knowing that a cloud provider can support a peer as part of some strategic alliances can help address users' last-minute requests.
We take into consideration the networks that are built upon the three components of a business process namely social network of persons, support network of machines, and configuration network of tasks. These networks are used to establish the coupling level between tasks, the partnership levels between machines, etc. In addition to these networks we capitalize on CloudPrO in our previous work which provide semantic and formal definitions for cloud resources in business processes.
Actually, there is a common consensus on the role that social technologies could play in improving business process management. However despite research efforts to incorporate social tools into processes, the contributions remained not tangible especially for cloud-based business processes. In this work we discuss a specific challenge, which is the lack of standardization to describe resources used by business processes deployed on the cloud. Building upon our previous work on social business processes we develop a framework We aim to ensure a free-of-conflict resource allocation during the process development.
As illustrated in Fig. 1 we express SoSeC-BPO in RDF/RDFS so that we manage it as a knowledge base upon which SPARQL queries are run in order to define conflict resolution strategies. The knowledge base is continuously updated whenever a business process is re-engineered or resources are re-assigned.

Framework Overview
Fig1. Framework Overview

To this end, we develop a Social Semantic for Cloud-based Business Processes Ontology (SoSeC-BPO) that supports free-of-conflict resource allocation. This ontology enables to standardize the social relations between a business process components.

Resource Conflict Resolution:
In the case of resource or task constraint violation, conflicts hampering business process completion can occur. To address these conflicts, our framework suggests strategies defined as semantic rules. According to our ontology, a conflict can be defined for a set of resources and handled by a strategy (Conflict Strategy) that consists of actions. Thus, we specify semantic rules following (E)vent-(C)ondition-(A)ction structure (On Event If Conditions Do Actions). In our work, events represent conflicts and conditions denote constraints. Actions suggests the strategies which are a set of solutions to take for resolving the conflict in question. These actions may include the elasticity feature of cloud by increasing resources capacities, or make additional resources or tasks. Many solutions can exist for the same conflict. To formally describe our ECA-based conflict resolution strategies, we use Semantic Web Rule Language (SWRL) in conjunction with Web Ontology Language (OWL) and Rule Markup Language (RuleML). We present in Table 1 the SWRL descriptions referring to our rules in accordance to the running example.

Table 1. SWRL Strategies for conflict resolution

We describe herein all strategies in detail:
  1. Strategies denoted as SSoSeC-BPO1 consist of:
    • Event: t11 need c3 which is non shareable and already consumed by t3.
    • Conditions: if there is a substitute resource c6 (backup(c6,c3)).
    • Actions: allocate c6 to t11 (consume(t11,c6)) or look for another substitute by comparing resources memories and speed.
  2. SSoSeC-BPO2 consists of:
    • Event: the storage resource st5 is non sufficient to respond to the requested memory of t4.
    • Conditions: if t4 is vertically elastic, otherwise, if it is horizontally elastic.
    • Actions: allocate st4 to t9 in addition to st5 or replicate t4 as well as st5.
  3. Strategies denoted as SSoSeC-BPO3 is composed of:
    • Event: the network resource net13, which is consumed by t13, become out of commission.
    • Conditions: if backup constraint is respected that means there is a substitute net10, or if t12 replaces t13 and t12 and st2 are in partnership.
    • Actions: assign r12 to t13 or t12 takes over and consume t12 and st2.
  4. SSoSeC-BPO4 Strategies consist of:
    • Event: when t6 requests more than capacity of c5 which is limited.
    • Conditions: if c1 can serve as backup.
    • Actions: c1 will be allocated to t6.
  5. Strategies denoted as SSoSeC-BPO5 comprises:
    • Event: if there is an unavailability of a human resource h2 and t2 is consuming it.
    • Conditions: If the Delegate constraint is fulfilled, otherwise if t2 supervises t1 work.
    • Actions: h1 takes over the adequate assignment. Otherwise, the t1 pursues the work.

Framework validation:
Proof of concept: To validate our framework, we extend the Signavio process editor, which is an open source web-based application and a powerful tool for mastering process management, with a semantic layer implementing our semantic framework and conduct a real use case study from France Telecom/Orange Labs. The plugin takes as input a selected activity from a BPMN process with the aim to attach a resource that it consumes. Then, a mapping with the SoSeC-BPO ontology is realized and a business process model annotated with the set of consumed resources is generated as output. Resources are described with attributes, properties and if exist its relations with another resources.

How To Use:
  1. The Signavio Process tool allows to design a business process model. So we design our example which is a real use case from France Telecom/ Orange labs.

  2. The user selects the appropriate activity to which he needs to attach necessary resources to consume by simply clicking on this task. Through the use of the SoSeC-BPO button in the south panel, the user have the possibility to access to the SoSeC-BPO ontology

  3. Once the button activated, the same panel shows all the available resource instances in the SoSeC-BPO ontology in order to enable the user to select the suited instance with appropriate attributes values to link which meet requirements.

  4. Finally, an extended business process model is constructed and annotated with required cloud and human resources. Consequently, data attributes like memory for storage resources or speed for compute resources are outlined within the text annotation. Besides, if there exist relations between resources, it is defined through last lines in the text annotation. So all knowledge about resources become more clear and explicit for the process owner.

Authors: Emna Hachicha (emna.hachicha@telecom-sudparis.eu) - Walid Gaaloul (walid.gaaloul@telecom-sudparis.eu) - Zakaria Maamar (zakaria.maamar@zu.ac.ae)

Last update on 01/10/2015