It would be nice to determine stall rates and the effects of truncation analytically, but these effects are dependent on workloads and also connectivity among clients and servers, and are not amenable to analysis. Instead, we study the effects via simulation. This section describes the simulation experiments and results. The results show that the stall rate is low even in workloads that stress the system, that truncation keeps multistamps small with little impact on stall rate, and that aging alone is not sufficient for truncation in these workloads. The section also discusses the cost of supporting causality and presents an optimization that further reduces stalls.