Although migration and replication of applications in a distributed environment have been discussed by many researchers, the implementations of these features are rarely focused when deployed in the Cloud. Distributed computing started with the cluster, evolved into the grid and now in the era of Cloud computing, where services are provided on demand and used with a pay-as-you-go style. In the Cloud, the enterprises usually have to migrate or replicate partly or fully their services because of economical or disaster preventing reasons. Because the cost of copying/moving the whole virtual machines is too high due to their big size, the replication/migration at application level is a possible approach. This work proposes an autonomically fine-grained framework for migrating/replicating the service components on the Clouds, which is a multi-cloud distributed deployment platform. Our proposed framework ensures the high availability and scalability of services, and complies with the service-oriented architecture. Our experiments conducted in real scenarios of elastic demands show that the proposed fine-grained migration/replication is more efficient than the coarse-grained ones when an autonomic system responds to fluctuation of webapp's workload.