Skip to main content

Centralized Security Data Ingestion

The reason to run one platform instead of six tools is not the dashboard; it is that the tools' outputs become one model — assets, findings, alerts, risks, controls, evidence — with identity carried across sources. This page says, source by source, what actually converges, so that "single pane of glass" is a description and not a slogan.

What each source contributes

SourceAssetsFindingsAlertsEvidence
Cloud scans (AWS, Azure, GCP)Every discovered resource, with identity and network contextMisconfigurations as cloud findings → unified findingsNew / reopened critical and high findingsControl evidence auto-collected per framework
Kubernetes and container scansClusters, workloads, images, registriesCluster misconfigurations, image CVEs → unified findingsCritical / highContainer and K8s compliance reports
Code scans (SAST, secrets, SCA, IaC)RepositoriesCode findings → unified findings; SBOMs → AIBOMCritical / highSBOMs, licence notices
Web / API / TLS / network scans (public or private)Scan findings → unified findingsCritical / highPer-scan reports
WazuhEndpoints, correlated with their cloud VMHost CVEs → vulnerability occurrencesRule-level ≥ 7 detectionsSCA / FIM browsed (not stored)
Threat feedsEnrich findings (KEV, EPSS, indicators)
Greenbone / OpenVAS, catalog-only toolsNot imported today

What "unified" means, concretely

  • One inventory. Asset Inventory holds cloud resources, Kubernetes clusters, repositories and Wazuh endpoints under one identity scheme; a Wazuh agent on an EC2 instance is the same asset the cloud scan discovered, not a duplicate. The security graph behind Attack Paths is built from the same records.
  • One findings lake. Every scanner writes into the unified findings store with a fingerprint (source, check, resource), so a re-scan updates a finding instead of creating another, a finding that disappears is reconciled as resolved, and the same CVE on an image and on a host are two occurrences of one vulnerability. Triage scores across all of them with one formula — severity, KEV, EPSS, exposure, environment, age.
  • One alert stream. Cloud, container, code, DAST and Wazuh events all pass through the same record_alert chokepoint into Alerts, de-duplicated with occurrence counts, routed by the same notification rules and SLAs.
  • One risk register and one evidence store. Findings from any source promote into the Risk Register; compliance evidence from cloud, container and Kubernetes scans lands in the Evidence Hub against SCF controls, so a framework score reflects the whole estate the platform can see.

Where correlation happens automatically — and where it does not

Automatic todayStill a person's job
Wazuh endpoint ↔ cloud VM (hostname, then IP)Wazuh host CVE ↔ OpenVAS result for the same host (OpenVAS is not imported)
Same finding across scans (fingerprint), same CVE across images and hosts (occurrences)Same application across a code repository, its container image and its running workload — linked where names match, not asserted
KEV / EPSS / indicator enrichment on every findingReading a Wazuh SCA failure as evidence for a specific control (browse, then attach)
Finding ↔ alert ↔ SLA ↔ ticket lifecycleBusiness context (owner, criticality) — set on assets by you, then used by SLAs and risk multipliers

Data residency

On an on-premises install every store above is yours: MongoDB and Redis on your host, evidence and reports in your object storage, scanners on your network. Nothing about your findings leaves unless you configure an outbound destination — a Slack channel, a Jira project, a webhook, an LLM provider — and those are per-team choices you can see in Integrations. The feeds and vulnerability databases the platform pulls carry nothing about you.