Compact deterministic distributed dictionaries (extended abstract)

Self-optimizing distributed trees

“…Most of these proposals explicitly balance the tree with each insertion and deletion, incurring a high cost for these operations. Distributed implementations of some of these algorithms (but not splaying) have also been proposed (e.g., [11,16,21,14]). Our approach is different in that it is less sensitive to insertions and deletions of nodes, and more sensitive to the actual workload.…”

“…In addition, several distributed implementations of important primitives such as mutual exclusion (e.g., [22,5,13,28,20]), and binary search (e.g., [11,16,21]) are built using tree structures, and these primitives are employed in a large number of distributed applications. The use of trees in different applications is warranted due to requirements specific to each setting, e.g., some applications map naturally to the tree hierarchy [17], others require the acyclicity of trees for simple protocol design [22,5], and yet others utilize the locality-preserving properties of trees [31].…”

A distributed data-balanced dictionary based on the B-link tree

“…Peleg [26,27] has studied the issue of data-balancing in distributed dictionaries from a complexity point of view, requiring that no pioce.t'or store more than O(M/N) keys, where M is the number of keys and N is the number of processors. In practice, this definition is simultaneously too strong and too weak, because it ignores constants and node capacities.…”