It is important to understand the volume of metrics that you are ingesting into Sumo Logic.
For billing and reporting purposes, data volume for metrics is measured in Data Points per Minute (DPM). When the DPM limit is exceeded, data is cached on the host and the Source is throttled. See the Account Page for information on how to check your current account usage relative to your available DPM limit.
The calculation of DPM varies according to the type of metric Source. See Metrics Sources for information about the available metric Sources.
The examples in this topic assume a one-minute scan interval, but shorter scan intervals will increase the DPM. For example, collecting at a 10-second scan interval will lead to six times the DPM of a one-minute scan interval.
Host metrics Sources
For host metrics, DPM depends on the number of hosts, the number of unique data points, and the scan interval. See Collected Metrics for a list of the specific host metrics that you can collect.
Disk metrics. Approximately 10 metrics are collected for each Source disk on each host.
Total disk metrics for single host = 10 x number of disks
Total disk metrics for multiple hosts = 10 x number of disks per host x number of hosts
Example: A host with 8 disks will generate 80 metrics. If you have 3 hosts, each with 8 disks, 10 x 8 x 3 = 240 individual metrics will be generated.
Network metrics. Network metrics are calculated per interface on each host, and approximately 4 metrics per interface are collected.
Total network metrics for single host = 4 x number of interfaces
Total network metrics for multiple hosts = 4 x number of interfaces per host x number of hosts
Example: A host with 6 network interfaces will generate 24 metrics. If you have 5 identical hosts, 4 x 6 x 5 = 120 metrics will be generated.
CPU, memory, and TCP metrics. Approximately 10 CPU, memory, and TCP metrics are collected for each host.
Total metrics for single host = 10
Total metrics for single host = 10 x number of hosts
Some Amazon namespaces report metrics every minute, while others report every five minutes. For each metric type, five individual metrics are reported (average, sum, samplecount, maximum, and minimum). The number of reported metric types varies according to namespace (a good general guideline for estimation is 10 per namespace instance).
Total CloudWatch metrics per namespace = 10 x 5
See the AWS documentation for details on the metrics for different namespaces.
Graphite Sources are custom, as are the number of collected metrics. Customers can specify the reporting interval external to the Sumo Logic configuration. The reporting interval determines when the Source sends data to Sumo Logic (which does not poll for the data). Be aware of this when you’re setting up the Source.