Abstract: Example implementations relate to clusters of nodes. In an example, a method includes computing a score for a first node in a distributed storage system comprising a plurality of nodes. The score is a function of at least one of: a location of the first node, a health status of the first node, and a resource utilization of the first node. A rank of the first node among the plurality of nodes is then determined based on the score and on a plurality of scores associated with the plurality of nodes. An instruction is sent to the first node to instantiate a manager process for managing the distributed storage system on the first node when the rank of the first node is at least equal to a predefined rank.

Abstract: Example implementations relate to clusters of nodes. In an example, a method includes computing a score for a first node in a distributed storage system comprising a plurality of nodes. The score is a function of at least one of: a location of the first node, a health status of the first node, and a resource utilization of the first node. A rank of the first node among the plurality of nodes is then determined based on the score and on a plurality of scores associated with the plurality of nodes. An instruction is sent to the first node to instantiate a manager process for managing the distributed storage system on the first node when the rank of the first node is at least equal to a predefined rank.

Abstract: A method, system and computer program product for managing data replication for data groups stored in a first storage device. A polling interval, a maximum bandwidth and a bandwidth tolerance available for data replication is defined. A priority and a status for each data group is defined. The data replication is started in the polling interval, for the data group with highest priority in the pending status to a second storage device connected to the first storage. The rate of data transfer during a polling period is determined by dividing the total data transferred during the polling interval by time period of the polling interval; and bandwidth utilization is determined for data replication by comparing rate of data transfer with maximum bandwidth. If the bandwidth utilization is less than the maximum bandwidth available then another data group is selected for replication. If the data bandwidth utilization is more than the maximum bandwidth available then selected data groups replicating are paused.