Described herein is a computer implemented method for managing summary data maintained for each node in a tree-structured dataset. The method comprises initiating a summary data reset process which comprises: determining a new summary data length; recalculating data summaries for all nodes in the tree-structured dataset so that data summaries for all nodes have a length is equal to the new summary data length, and wherein for a given subtree of the tree-structured dataset, recalculating data summaries comprises: initially recalculating data summary items for leaf nodes of the given subtree; and following recalculation of data summary items for leaf nodes, working progressively up the given subtree to a root of the given subtree.