NVMe over Fabrics Scale-Out Management

In this video from the 2019 OpenFabrics Workshop in Austin, Phil Cayton from Intel presents: NVMe over Fabrics Scale-Out Management.

"Los Alamos National Laboratory's in-house high speed network monitoring software was developed to provide
information about fabric issues and useful information about fabric utilization. The software has successfully met these
objectives in the past, however new interest in increasing its capabilities has risen from multiple teams at LANL. The goal
of this project is to increase the frequency of gathering fabric performance counters such that more granular
understanding of network utilization is possible. Prior to modification, the execution frequency of the software could be
modified with the configuration settings. However, the run time of the monitoring code was greater than our objective
sample rate for our largest systems, so gathering fabric utilization data at the desired frequency was not possible. To
implement the enhancements, the code was modified so that fabric counter gathering was isolated to a separate
process with an improved algorithm."