> ## Documentation Index
> Fetch the complete documentation index at: https://docs.aviatrix.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Stop Data Exfiltration Through Default-Deny Egress Control

> How Aviatrix stops data exfiltration and command-and-control traffic with default-deny egress and managed threat feeds, covering the business case and key components.

Unrestricted outbound internet access is the primary channel for data
exfiltration and encrypted command-and-control traffic. Aviatrix closes that
channel by enforcing default-deny egress inside each VPC or VNet, where the
traffic originates, so a compromised workload cannot reach a destination you did
not approve.

## Business Challenge

Cloud workloads reach the internet by default, and a native NAT gateway forwards
every outbound connection while recording little about it. A compromised
workload or a poisoned dependency can send data to any destination, including
over encrypted channels that look like ordinary traffic. Each cloud also ships
its own egress controls with its own rule format and limits, so teams that run
more than one cloud maintain several policies for one goal. Without one place to
see which destinations each VPC or VNet can reach, you cannot tell how much of
your estate is covered, and you cannot show an auditor.

## The Business Case for Egress Control

### Risk Mitigation

* **Data exfiltration.** An outbound path that is closed by default removes the
  channel an attacker uses to move stolen data out of your estate.
* **Command-and-control disruption.** A workload that cannot reach an unapproved
  destination cannot receive instructions or download further tooling.
* **Supply-chain exposure.** A poisoned dependency can call only the
  destinations that you approved.
* **Known-bad destinations blocked.** Managed threat and country feeds stop
  traffic to destinations that are already known to be malicious.
* **Audit evidence.** Enforced outbound rules give an auditor a control to
  inspect instead of an assurance to accept.

### Operational Efficiency

* **One policy model across clouds.** Teams maintain one approach to egress
  instead of one rule format for each cloud.
* **Domain-based allow-lists.** Rules reference destination domains rather than
  IP addresses, so they stay valid when a SaaS provider changes its addresses.
* **Allow-lists built from observed traffic.** Teams permit the domains that
  workloads actually call instead of guessing.
* **Feeds that stay current.** Aviatrix maintains the threat and country feeds,
  so your team does not curate address lists.
* **Reversible rollout.** Teams monitor first, protect when the traffic is
  understood, and return a VPC or VNet to monitoring if an application breaks.

### Strategic Value

* **Measurable posture.** A score for each VPC or VNet lets leaders prioritize
  work and report progress.
* **Egress at the workload boundary.** Inspection in the VPC or VNet where
  traffic originates replaces forwarding every connection through a native NAT
  gateway.
* **A reusable policy model.** The same groups and rules that control egress
  also control east-west traffic, so the investment carries forward.

## How Aviatrix Addresses Egress Risk

Aviatrix Distributed Cloud Firewall (DCF) with WebGroups is the recommended way
to secure egress. DCF evaluates traffic at the Spoke gateway in the VPC or VNet
where it originates, before it leaves the cloud network.

![Egress security architecture: workloads in a VPC or VNet send outbound traffic to an Aviatrix Spoke Gateway deployed in the same VPC or VNet and acting as the Policy Enforcement Point. It runs Distributed Cloud Firewall with local egress and SNAT, permits domains in approved WebGroups, blocks everything else and logs each deny, and reports posture as the Egress Security Score in CoPilot.](https://mintcdn.com/aviatrix-14b37c43/fGohijthCmM99wVL/images/solutions/stop-data-exfiltration/egress-containment-architecture.png?fit=max\&auto=format\&n=fGohijthCmM99wVL\&q=85\&s=5b3f1d76644a45c85268b82874f12658)

*A Spoke gateway inside the VPC or VNet permits approved domains, blocks and
logs everything else, and feeds the Egress Security Score.*

### Roll Out in Stages

You move each VPC or VNet from monitoring to protection instead of enforcing
everywhere at once. Monitoring shows where workloads connect before any traffic
is blocked. Protection permits the trusted domains that you select and denies
the rest. If an application breaks, you return the VPC or VNet to monitoring
without losing visibility.

### Allow-List by Domain

WebGroups group the domains that workloads are allowed to reach, so a rule
describes a business destination instead of a changing set of addresses.

### Block Known-Bad Destinations

ExternalGroups use Aviatrix-managed feeds: a threat feed of malicious addresses,
a countries feed for geographic policy, and a feed of SaaS services. You
reference them in the same rules as SmartGroups and WebGroups. This model
replaces the deprecated ThreatIQ and Geoblocking features. You cannot use
ThreatIQ or Geoblocking together with DCF and ExternalGroups, so plan the
migration as one change. The migration requires Controller 8.1 or later.

### Inspect Encrypted Traffic Where Policy Requires It

URL-based WebGroups and intrusion prevention (IPS) signatures need visibility
into encrypted traffic. A DCF rule can enable TLS decryption for only the flows
that need it. You attach an IPS profile to the rules that need inspection, so
inspection applies only to the traffic that you select.

> [!NOTE]
>
> DCF intrusion prevention is generally available in Controller 9.0 and later.
> In Controller 8.2, it is a Preview feature.

### Measure Your Posture

The Egress Security Score rates each VPC or VNet from 0 to 1. A score of 0 means
traffic is neither blocked nor logged, and a score of 1 means default deny with
all outbound traffic logged. CoPilot shows how many VPCs and VNets sit at each
protection level.

## Key Components

> [card] [Distributed Cloud Firewall](https://docs.aviatrix.com/docs/enterprise/9.0/concepts-architectures/components/dcf/dcf-overview.md)
>
> Enforces default-deny egress rules at the Spoke gateway in each VPC or VNet.

> [card] [WebGroups](https://docs.aviatrix.com/docs/enterprise/9.0/concepts-architectures/components/dcf/webgroups-about.md)
>
> Keeps egress rules valid by allow-listing domains instead of IP addresses.

> [card] [SmartGroups](https://docs.aviatrix.com/docs/enterprise/9.0/concepts-architectures/components/dcf/smartgroups-about.md)
>
> Applies egress policy to the VPCs, VNets, and workloads you choose.

> [card] [Egress Security Score](https://docs.aviatrix.com/docs/enterprise/9.0/concepts-architectures/components/egress/egress-overview.md)
>
> Shows how well each VPC or VNet is protected, in CoPilot.

> [card] [ExternalGroups and Feeds](https://docs.aviatrix.com/docs/enterprise/9.0/reference/externalgroup-about.md)
>
> Aviatrix-managed threat, country, and SaaS feeds that you use in rules.

## Related Documentation

Use these documents to plan and implement egress control.

### Configuration Guides

* [Implementing Egress in an Aviatrix-Managed Network](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/egress/egress-introduction.md)
  — prerequisites and the overall workflow.
* [Monitoring Egress Traffic](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/egress/egress-monitor-vpc-vnets.md)
  — see where workloads connect before you block anything.
* [Protecting Egress Traffic](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/egress/egress-protect-vpc-vnets.md)
  — permit trusted domains and deny the rest.
* [Unprotecting and Monitoring a VPC/VNet](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/egress/egress-unprotect-monitor.md)
  — return to monitoring while you troubleshoot.
* [Enabling Local Egress](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/egress/egress-enabling-local.md)
  — route a VPC or VNet through its Spoke gateway.
* [Aviatrix Cloud Firewall Secure Egress Implementation Guide](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/egress/aviatrix-cloud-firewall-implementation.md)
  — network design and end-to-end implementation.
* [Migrating from Legacy Egress to Distributed Cloud Firewall](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/dcf/egress-migration.md)
  — move Egress FQDN filtering to DCF and WebGroups. The migration requires
  Controller 8.1 or later.
* [Migrating ThreatIQ and Geoblocking to Distributed Cloud Firewall](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/threatiq-geo-dcf-migration-guidelines.md)
  — replace the legacy features with ExternalGroups.
* [Intrusion Prevention](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/dcf/ips.md)
  — enable IPS and attach profiles to DCF rules. IPS is generally available in
  Controller 9.0 and later, and is a Preview feature in Controller 8.2.
* [Distributed Cloud Firewall Settings](https://docs.aviatrix.com/docs/enterprise/9.0/guides/security/dcf/dcf-settings.md)
  — configure global DCF behavior, including decryption and IPS settings.

### Reference

* [About ExternalGroups](https://docs.aviatrix.com/docs/enterprise/9.0/reference/externalgroup-about.md)
  — feed types and behavior.

* [Egress Security Score](https://docs.aviatrix.com/docs/enterprise/9.0/reference/egress-filtering/egress-security-score.md)
  — how each score is calculated.

* [Distributed Cloud Firewall Capabilities](https://docs.aviatrix.com/docs/enterprise/9.0/reference/dcf/dcf-capabilities.md)
  — supported capabilities by Controller version.

* [Terraform Provider](https://docs.aviatrix.com/docs/enterprise/9.0/reference/terraform/provider.md)
  — declare egress policy as code.

For Aviatrix Cloud, see
[Egress Security Overview](https://docs.aviatrix.com/docs/cloud/security/egress-nat/about-egress-security.md)
and
[Configure TLS Decryption](https://docs.aviatrix.com/docs/cloud/security/dcf/tls-decryption-configure.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.