Cloudcomputing provides all types of resources as services. Hardware and softwareresources are shared with reference to the user demands. Service components aredeployed to support the enterprise applications. Virtual Machines (VM) areprovided for the service components based on the capacity levels. Servicecomponent consolidation or VM consolidation refers the process of mapping theVirtual Machines to the service components. The service components associatedwith the same application is denoted as Dependant Service Component.
Intelligentservice management models identify the service dependencies to achieve the lownetwork latency. The service consolidation with low communication time reducesthe communication cost for the network applications.Servicecomponents for enterprise applications are deployed in many hosts and networkdevices. Service components consolidation is carried out with the considerationof resource constraints, service dependency and network structure. CloudScoutis a non-intrusive approach applied for automatically discovers dependentservice components. The correlation among service components is analyzed withthe time-series information from system monitoring logs. The log miningapproach is used for the service dependency discovery with generality andprivacy protection features.
The iEntropy method is applied to dynamicallydetermine the weight of each resource metric in the service distancecalculation. The hierarchical and iterative k-means (HiKM) algorithm is adaptedfor the dependent service clustering process. The Queue network and networklatency optimization are used in the VM consolidation process. TheCloudScout scheme is enhanced to support optimal service dependency discoveryand service consolidation process. The service dependency discovery process isimproved with sequential dependencies and complimentary dependencies. The MultiObjective Service Consolidation (MOSC) scheme is build to allocate the VirtualMachines for the service components. The service consolidation process isupgraded with resource level constraints.