Cross-device alert and log correlation

What it does

Correlation turns many related alerts into fewer things to work on. A device's alerts that open a case (the Case column on Alert Rules) share that device's case. When the same check fails on 3 or more devices of one customer at the same time, those alerts become one storm incident. When a run of collected log events matches a correlation rule, you get one security finding. Correlation never crosses customers or workspaces.

How it works

  • Cases per device. A new alert joins the open case it belongs to: first the case for the same device, then, for alerts without a device, the case for the same customer and the same item. A case takes the most severe level of the alerts in it and never becomes calmer by absorbing an alert. A case resolves on its own only when none of its alerts is still open.
  • Storm incidents. When the same check is failing on 3 or more devices of one customer at the same time (for example, a site outage taking many devices offline), the alerts are grouped into one incident. You get one notification when the incident opens and one when the last device recovers. Alerts that join while the incident is open do not notify on their own.
  • Security log correlation (native SIEM preview). Collected security events for a customer are checked against correlation rules, for example repeated authentication failures from one source, an audit or event log being cleared, a privileged account being created, a firewall deny sweep, security tooling being disabled, or a malware detection. A match opens a Correlated security finding alert, which opens a case by default. This is part of the native SIEM preview: on the Alert Rules page, the Native SIEM tab reads "Off unless the native SIEM preview is switched on." Ask support whether the preview is on for your workspace.
  • RevEngine Sentry. Sentry proposes possible relationships between one customer's open alerts, with cited root-cause hypotheses, for an administrator to evaluate. It does not group, route, suppress, close or contain alerts on its own, and correlation never authorizes containment or account disabling.

Step by step: set up

  1. Go to Administration > General > Alert Rules (Owner or Admin). Check the Correlated security finding alert kind: its severity, and whether it opens a case, notifies or creates a ticket.
  2. Set your notification routes as described in the Alerts, cases and notifications guide. Storm incidents use the same routes.
  3. Optional: go to Administration > RevEngine > Sentry (Owner or Admin). Under Workspace controls, set Sentry for this workspace to Enabled (shadow only) and select Save setting, then choose a customer and select Run Sentry shadow investigation. Results appear under Recent shadow investigations. Sentry is off by default, and runs also need the deployment gate to be on.

Day to day

  • Work the case or the incident, not each alert. The case lists the alerts in it under Correlated signals, each with its own severity.
  • When the last member of a storm incident clears, the incident resolves and sends one notice.
  • At least two open alerts are needed for a Sentry investigation.

Limits

  • Correlation is always within one customer of one workspace. Cases never span devices; only storm incidents do.
  • Storm grouping needs the same check on at least 3 devices of the same customer at the same time.
  • A Sentry run you start looks at the latest 20 open alerts of the customer. When there are more, it says "Sample limit reached. Other open alerts were omitted; no complete coverage is claimed." Background relationship evaluation inspects up to 256 open alerts per customer in separate windows of at most 20 alerts; larger windows that cannot safely be separated remain inconclusive.