This dataset contains the outcome of the MECPerf experiment, funded within the 6th Fed4FIRE+ open call for experimenters. MECPerf is an open source tool to measure key performance indices (KPIs) of a MEC/cloud infrastructure, at both network and application level. MECPerf is available at https://github.com/MECPerf/MECPerf. The dataset contains bandwidth and latency measurements collected in three different operational scenarios: i) active measurements, in which MECPerf measures KPIs such as TCP bandwidth, UDP bottleneck capacity, TCP and UDP RTT of a MEC and a cloud environment, in presence of increasing cross traffic; ii) passive measurements through edge application, in which a DASH streaming application running on multiple clients measures its own KPIs in presence of increasing cross traffic, and communicates them to MECPerf via an open interface; iii) passive measurements through traffic redirection, in which MECPerf sniffs the traffic produced in the same setup as scenario ii to extract KPIs. More details can be found in the deliverables and final report of the MECPerf project.