Building A Layer-2 Hybrid Cloud

Cloud computing enables the deployment of customer or enterprise services on a specific cloud in a pay-as- you-go fashion, with the flexibility to scale resources elastically. As cloud computing technologies gain popularity, enterprises are increasingly transitioning some of their services from local physical servers in data centers to cloud environments, including both public and private clouds. Due to varying privacy requirements and the diverse capabilities of computing, storage, and networking, enterprises often distribute their services across local data centers, public clouds, and private clouds. For instance, a company's private service would be deployed in a local data center, an open service catering to the public could be hosted on a public cloud with horizontal scalability, while other internal services may be deployed on a private cloud operated by the company itself. These services often require communication among themselves and may need to be migrated from one site to another while preserving their IP addresses. For example, a private service may be migrated from a local data center to a public cloud while maintaining accessibility from both the local data center and the company's private cloud. Traditional hybrid cloud approaches typically rely on layer-3 methods, routing, and tunneling traffic between different sites. However, these routing-based hybrid schemes cannot support IP-stable migration scenarios because they require each site to have a unique IP prefix. In this paper, we propose an SDN- enabled, layer-2-based implementation of a hybrid cloud. This approach establishes connectivity between resource pools located on different sites and virtual private clouds (VPCs) within the same site. It facilitates IP-stable service migrations between different sites, enabling seamless communication and preserving IP addresses during the migration process.