A trust-based, multi-agent architecture supporting inter-cloud VM migration in IaaS Federations

The success of inter-cloud VM migration depends on several different factors for which a good estimate should be made in advance by Cloud providers. In this context, the ability of the counterparts in performing related measures and appropriate evaluations will affect the success of the migration to be performed. We argue that the risks due to errors made in the evaluations above can be classified into two different classes: i) a direct (negative) consequence on the reputation built towards the client customer, and ii) additional costs due to waste of resources. In this work we propose a trust-based, multi-agent architecture on which software agents assist providers in taking decisions about VM migration, by performing an evaluation of the different risks discussed above. We envision a first class of software agents that computes trust information by means of a model that considers the critical concerns to take into account in advance for the VM migration process. A second class of software agents is employed to assist the decisional agents in per- forming measures about VMs and measuring errors in front of previous VMs migrations. Moreover, they are also employed to disseminate trust information by means of a gossip-based protocol, in order to provide fault tolerance, efficiency and redundancy.