Online Social Networks (OSNs) are becoming more and more popular in today’s Internet. Distributed Online Social Networks (DOSNs), are OSNs which do not exploit a central server for storing users’ data and enable users to have more control on their profile content, ensuring a higher level of privacy. The main challenge of DOSNs comes from guaranteeing availability of the data when the data owner is offline. In this paper we propose a new P2P dynamic approach to the problem of data persistence in DOSNs. By following Dunbar’s approach, our system stores the data of a user only on a restricted number of friends which have regular contacts with him/her. Users in this set are chosen by considering several criteria targeting different goals. Differently from other approaches, nodes chosen to keep data replicas are not statically defined but dynamically change according to users’ churn. Our dynamic friend selection achieves availability higher than 90% with a maximum of 2 online profile replicas at a time for users with at least 40 friends. By using real Facebook data traces we prove that our approach offers high availability even when the online time of users is low.