Azure Autobalanced Collectors - 5 minute windows of ping failures

When you look at the Collector device that is monitoring the devices where the PING loss is happening, I would recommend checking the Collector Data Collecting Tasks datasource to see if there is some sort of cyclical overload of those tasks happening. Specifically I would look at the graphs for Unavailable Thread Scheduling and Queue datapoints before, during and after the Ping loss time periods. It may indicate that the collector ABCG is not keeping up with Ping specifically, which is something you may be able to tune around.