Research On Evolution Technologies Of Internetware Based On Semantic Relationship

Internetware is different software entities,which use of a service as basic structural units,distributed application support interconnected.In response to the challenges for application development,deployment,running and maintenance in an open,dynamic,difficult to control for Internet platform,the concept internetware came into being.Internetware evolution has static evolution and dynamic evolution dynamic.The research of static evolution should start with the software architecture(SA)level,which set the changing of component to the SA influence.And dynamic evolution is able to ascertains whether run-instance of software can executes migrate from flow model(source model)to new flow model(target model).Now,the mainly static evolution of research is lack of quantitative analysis the influence scope of SA to varied component.While,the research of dynamic evolution is data flow depend on control flow,and then it can not meet the requirement of internetware control flow separate from data flow.Based on the above characteristics of intemetware,this paper makes research in the following aspects.(1)The static evolution characteristics of intemetware is analyzed.Firstly,components,semantic relationship between components and their characteristics of internetware architecture is studied.Based on semantic protocol,the sequentiality and logicality of semantic relationships interior component and between components is analyzed,the sequentiality and logicality of semantic relationships of components is expressed by semantic protocol item based on semantic relationship.The semantic relationships matrix and semantics relationship-link matrix of internetware architecture are built,and their characteristics is analyzed.On this basis,internetware of component adding,modifying,deleting,merging,splitting operation is studied,the quantitative index of effect on SA evolution operation is presented,and quantitative index of effect on static evolution operations is presented.(2)Ripple-effect of intemetware static evolution based on semantics relationship is proposed.According to the semantic relationship characteristics of intemetware architecture and static evolution operation characteristics,a formal description of internetware semantics which can describes component port semantics and between ports by method invocations is built.Architecture formal model of internetware-oriented static evolution characteristics is proposed.Based on semantics relationship matrix and it's quantitative of internetware architecture,the quantitative analysis is proposed to measure ripple-effect index of semantics relationship chain and semantics relationship component of static evolution operation range is put forward.According to the degree of semantics protocol relationship,correspondence effect and degree of different evolution operation with adding,modifying,deleting,merging and splitting of component is studied.Ripple-effect index of semantics relationship chain and ripple-effect index algorithm of semantics relationship component of different evolution operation are raised.(3)Dynamic evolution error of data flow and control separation about internetware is analyzed.Firstly,a data flow and control flow oriented internetware-sevice model based on colored Petri Net is put forward,it is introducing the data flow explicitly into internetware service modeling process.Secondly,during interrnetware serving,data dependence relationship and its characteristics is researched for static evolution and dynamic evolution.It appears data flow and control flow have equal importance of internetware service dynamic adaptation and evolution.Based on this,it analyzes data dependence relationship of internetware service model.To emphasize keep data dependence relationship during dynamic evolution.Next,by means of internetware service model of data flow and control flow oriented,dynamic evolution operation pattern and features are analyzed.The changing of data flow constraints,data flow and control flow cross dependence during internetware service dynamic evolution process,which may lead to mistake features of data flow and control flow.(4)Dynamic migratability criterion of internetware service instances is proposed.First,based on internetware service model of data flow and control flow oriented,according to data dependence characteristics of intemetware service model and dynamic evolution of possible faults after the separation of data flow and control flow.Dynamic migratability criterion of data flow oriented internetware service instance is established.Not only between source mode and target mode data dependency relationship is considered,parameters of non-repetitive reappearance between source mode and target mode was also be considered.A generating algorithm about data dependence according to migratability criterion is designed.In terms of data flow and control flow cross interaction characteristics after separating,internetware service instance about data/control cross dependence of migratability criterion is raised,and prove this dynamic service intance target state of migratability criterion is effective,which depict the dynamic migration constraint characteristics of complete internetware service instance.