MultiGranular: An effective Service Composition Infrastructure for Multi-tenant Service Composition

service Figure 1. The Architecture of MultiGranular The architecture of MultiGranular is shown in Figure 1. In order to respond to the complexity of multi-tenant service composition in cloud computing, we give the corresponding solving strategies: Hierarchy. Considering from similar, correlation and aggregation relationships of the granularity among services, the process of MultiGranular computing is introduced, which includes basic service clustering, correlation mining and hierarchy building, which makes volumes of disorder services into the hierarchy and ordered clustering services, effectively responding to multi-granularity service composition requirements. Uncertain correlation. Uncertain of correlation relationships are measured from the aspect of correlation degree between services. The computing of correlation degree is based on matching between concepts of services message and similarity measurement. Inheritance. Tenants can firstly build sub service reflecting their individual requirement by inheriting on the mostly similar service with their requirements with least inheritance operation. Then corresponding concrete service will be got from which the inherited service maps to for customization. Inheritance is realized by modifying attributes of services, such as UI, business process, descriptive definition and data schema to satisfy tenants’ individual requirements. In section 5 we will introduce some inheritance operations. Versioning. The frequent evolution of services to meet business requirements changed always leads to many versions of the same original service. Therefore, it is necessary to realize version management mechanism. On l n e V ers ion O nly . Bo ok m ad e b y t his fil e i s I LL EG AL . International Journal of Multimedia and Ubiquitous Engineering Vol.9, No.6 (2014) 174 Copyright c 2014 SERSC 4. MultiGranular Computing As Figure 1 shows, MultiGranular includes two hierarchical models, in which abstract services maps to concrete services mutually. Formally, it is represented as a quintuple, (S, Rs ,Rc, Ra, Ri, De), where: Definition 1. Service S (v) = (UI, Business process, descriptive definition, data) UI provides the uniform entrance for accessing service and describes the functional module and layout. Business process is the specific process for realizing particular goal and includes atomic process and composite process. It is consisted of services and flow relationships between them. Descriptive definition mainly declares the functionalities, such as input and output message and their data types. Data schema is consisted of a set of data items. v is the version number. Definition 2. Similar relationship Rs. Services that can realize similar functionality because of similar semantic concepts are similar. Similar relationship between services is defined as i j S S  , ( , ) i j Similar S S   There are many methods to compute similarity degree between services. In this paper, we compute the similarity degree based on their distance in the ontology. 1 ( , )= ( , ) i j i j Similar S S dis c c , 1 1 1 1 ( , ) ( , ) 1 ( , ) ( + ) 2 n m m n j i i j i j j i i j Similar in in Similar in in Similar Input Input m n Max Max         1 1 1 1 ( , ) ( , ) 1 ( , ) ( + ) 2 t s s t j i i j i j j i i j Similar Out Out Similar Out Out Similar Output Output s t Max Max         1 2 ( , ) ( , )+ ( , ) i j i j i j Similar S S w Similar Input Input w Similar Output Output    In the MultiGranular, we firstly divide service set into several groups, services in each of which have relatively high similarity in functionality while services between different of which vary a lot. After basic services clustering, each service cluster can be represented as an abstract service through extracting the same concept respectively in the input and output message of services in the same cluster . Definition 3. Correlation relationship Rc. Correlation relationship is the basis of service composition. According to the matching relationship between concepts of input and output message in the ontology, two forms of correlation relationships between services are defined as follows: (1) c R i j S S  , if . . , ( . , . ) i j i j S Output S Input Similar S Output S Input    (2) 1 2 c R n j S S S S     ,if 1 2 . . . . , , ( , ) ( . , . ) n j i i j i j S Output S Output S Output S Input S CD S S Similar S Output S Input         In the second step of building MultiGranular, the two forms of correlation relationship will be mined among abstract services according to semantic matching. Definition 4. Aggregation relationship Ra. As coarse-grained services can hierarchically decompose to fine-grained services that have correlation relationship, we measure the two forms of aggregation relationship in the aspect that one service can realize composition function of services correlated. On lin e V ers ion O nly . Bo ok m ad e b th is f ile is IL LE GA L. International Journal of Multimedia and Ubiquitous Engineering Vol.9, No.6 (2014) Copyright c 2014 SERSC 175 (1) , a R i j k S S S  ,if . . , : , ( . , . ) . . , : , ( . , . ) c R i j