Inter-nodes multicasting communication in a monitoring infrastructure

A method comprising:

• collecting, by one or more processors on a first agent, a first instance of monitoring data relating to an activated alert on a computer system, wherein the collecting is based, at least in part, on a set of instructions received from a monitoring server, wherein the set of instructions identifies the first agent and a second agent of a plurality of agents in a multicast group, and wherein the second agent monitors for a second instance of monitoring data;
• obtaining, by one or more processors on the first agent, a value for a first metric that is included in the collected first instance of monitoring data;
• calculating, by one or more processors on the first agent, a delay period to wait before sending the first instance of monitoring data to the monitoring server and to the plurality of agents in the multicast group, wherein the delay period is based on how likely an alert condition is to be met based on the value of the first metric;
• receiving, by the one or more processors on the first agent, the second instance of monitoring data from the second agent;
• obtaining, by one or more processors on the first agent, a value for a second metric that is included in the received second instance of monitoring data;
• determining, by the one or more processors on the first agent, whether the collected first instance and the received second instance of monitoring data are relevant to triggering the activated alert based, at least in part, on whether the value for the first metric and the value for the second metric, are above respective alert thresholds;
• responsive to determining that the collected first instance and the received second instance of monitoring data are not relevant to triggering the activated alert, determining, by the one or more processors on the first agent, not to send the first instance and second instance of monitoring data to the monitoring server;
• and
• responsive to determining not to send the first instance and the second instance of monitoring data to the monitoring server, recalculating, by one or more processors on the first agent, the delay period, wherein the recalculated delay period is based on how likely an alert condition is to be met based on the values of the first metric and the second metric.