Automatic Failover of Nodes of a Middle-Tier Layer

Method and system are provided for automatic failover between multiple nodes in a middle-tier layer between client applications and a back-end service. The method at a first node includes: receiving a request from a client application; determining that the first node cannot service the request successfully; forwarding the request from the first node to a second peer node; receiving a response at the first node on completion of the request by the second peer node; and forwarding the response from the first node to the client application. An original path of communication between the client application and the first node is maintained. The failover of the request from the first node to the second peer node is transparent to the client application. Nodes may be registered in a group, wherein a node in a group has awareness of other nodes in the group.