In tree-based hierarchical key management schemes, scalability is achieved by reducing the number of messages exchanged during a rekeying operation. A single server manages the entire tree structure in such schemes. Failure of that server leads to single point failure which interrupts the group communication. In this paper we propose a method to avoid single point failure by distributing user information among set of X number of servers and use (t,X) threshold scheme to reconstruct the tree. The new auxiliary keys and group key are computed partly by the users which reduces number of encryptions required to communicate new set of keys to the remaining group members.