# Network Segmentation

Network segmentation divides a network into distinct zones and controls communication between them according to security requirements.

Cloud concept. Updated 2026-10-29. Page: https://blowstack.com/tools/cloud-concepts-hub/network-segmentation

> Network segmentation separates workloads by trust level, function, environment, or data sensitivity, then permits only the traffic required between them. It reduces exposure and limits lateral movement after a compromise, but effective segmentation requires enforced traffic controls rather than merely separate address ranges.

Practice exams covering Network Segmentation: [ANS-C01 AWS Certified Advanced Networking - Specialty](https://blowstack.com/practice-exams/aws-certified-advanced-networking-specialty), [CLF-C02 AWS Certified Cloud Practitioner](https://blowstack.com/practice-exams/aws-certified-cloud-practitioner), [SOA-C03 AWS Certified CloudOps Engineer - Associate](https://blowstack.com/practice-exams/aws-certified-cloudops-engineer-associate), [DOP-C02 AWS Certified DevOps Engineer - Professional](https://blowstack.com/practice-exams/aws-certified-devops-engineer-professional), [MLA-C02 (beta) AWS Certified Machine Learning Engineer - Associate](https://blowstack.com/practice-exams/aws-certified-machine-learning-engineer-associate), [SCS-C03 AWS Certified Security - Specialty](https://blowstack.com/practice-exams/aws-certified-security-specialty), [SAA-C03 AWS Certified Solutions Architect - Associate](https://blowstack.com/practice-exams/aws-certified-solutions-architect-associate), [SAP-C03 AWS Certified Solutions Architect - Professional](https://blowstack.com/practice-exams/aws-certified-solutions-architect-professional).

## 6 key aspects of Network Segmentation

Effective segmentation combines requirement-driven boundaries, least-privilege traffic rules, controlled routing, and ongoing validation.

1. **Security zones based on requirements**: Group workloads by trust level, business function, environment, or data sensitivity. Define required flows before choosing subnet layouts or connectivity services.
2. **Least-privilege communication**: Permit only necessary sources, destinations, protocols, and ports between zones. Separate subnets or address ranges do not provide this restriction without enforced controls.
3. **North/south and east/west protection**: Control traffic crossing environment boundaries as well as traffic moving between internal workloads. Internet-facing protections alone do not prevent lateral movement.
4. **Explicit routing and isolation**: Use route tables and gateways to establish intended paths and remove unintended connectivity. Distinguish direct internet access, outbound access through NAT, and isolated subnets with no internet route.
5. **Layered enforcement**: Combine workload-level filtering, subnet controls, and inspection where requirements justify them. Account for stateful versus stateless behavior and ensure traffic cannot bypass required inspection.
6. **Continuous validation**: Test that required connections succeed and prohibited connections fail. Review flow logs and configuration changes to detect overly broad permissions, routing errors, and unintended connectivity.

## Cloud services for Network Segmentation

Amazon VPC supplies the network boundaries and local traffic controls, while AWS Transit Gateway manages connectivity between multiple networks. Transit gateway route tables can keep groups of networks separated and direct selected flows toward an inspection VPC. AWS Network Firewall inspects traffic only when routing places its endpoints in the traffic path. AWS PrivateLink offers service-specific private connectivity without opening general network access between consumers and providers. Together, these services support layered segmentation, but their effectiveness depends on consistent routes, narrowly scoped permissions, and testing of both allowed and blocked communication.

- **[Amazon VPC](https://blowstack.com/tools/cloud-services-hub/amazon-vpc)**: Provides virtual networks, subnets, route tables, gateways, security groups, and network ACLs used to define and enforce network boundaries. Separate subnets still require appropriate filtering because local routing normally permits paths between VPC address ranges.
- **[AWS Network Firewall](https://blowstack.com/tools/cloud-services-hub/aws-network-firewall)**: Provides managed network inspection and filtering for traffic routed through firewall endpoints. It can protect north/south or east/west paths, but route design must prevent bypass and maintain appropriate forward and return paths.
- **[AWS Transit Gateway](https://blowstack.com/tools/cloud-services-hub/aws-transit-gateway)**: Connects VPCs and on-premises networks through a transit hub. Separate transit gateway route tables and controlled route propagation can restrict connectivity between groups of attachments and support centralized inspection architectures.
- **[AWS PrivateLink](https://blowstack.com/tools/cloud-services-hub/aws-privatelink)**: Provides private access to supported services without requiring broad network connectivity between consumers and providers. Interface endpoints can support service access from isolated subnets, subject to endpoint security groups, applicable endpoint policies, and service authorization.

## Network Segmentation exam questions

Real practice exam questions on Network Segmentation, with the correct answer and its explanation.

### Question 1 (CLF-C02)

**Question:**

A company runs workers that process files uploaded by external users and internal finance servers in separate private subnets of the same Amazon VPC. Current network controls allow all traffic between the subnets. The workers have no business need to connect to the finance servers. The company wants to prevent a compromised worker from initiating connections to those servers. Which change meets this requirement?

**Correct answer:** Replace the finance servers' permissive inbound security group rules with rules that allow only required connections from approved sources, excluding the workers.

**Explanation:** Restricting the finance servers' inbound rules enforces the boundary between the workloads. Excluding the workers from allowed sources prevents worker-initiated connections while preserving required access from approved sources.

### Question 2 (MLA-C02 (beta))

**Question:**

A company is deploying an Amazon SageMaker AI real-time inference endpoint. During inference, the custom model container must query a PostgreSQL feature database on TCP port 5432 in the company's VPC. Development instances also run in this VPC. Security requirements state that the model container must not initiate application connections to other VPC resources or the internet, and development instances must not be able to connect to the database. Model artifact access and IAM permissions are already configured. Which network configuration meets these requirements?

**Correct answer:** Configure the model's VpcConfig with subnets that have no internet gateway or NAT gateway routes. Attach only a dedicated model security group, replace its default outbound rule with TCP 5432 access to the database security group, and configure the database security group to allow inbound TCP 5432 only from the model security group. Leave SageMaker network isolation disabled.

**Explanation:** VpcConfig provides the model container with connectivity to the selected VPC. The dedicated security groups permit the required database connection while excluding development instances and other application destinations. Removing internet routes prevents internet access. SageMaker network isolation must remain disabled because enabling it would prevent the container from connecting to the database.

### Question 3 (SAP-C03)

**Question:**

A company runs frontend servers, processing workers, and PostgreSQL databases on Amazon EC2 instances in one VPC. Each Availability Zone has a private subnet that contains instances from all three tiers. Auto Scaling frequently replaces frontend servers and workers. The VPC has working routes to a NAT gateway and other corporate networks.

Every application instance has two security groups: a tier-specific security group and a shared baseline security group. The baseline group allows all inbound TCP traffic from itself, allows inbound TCP port 9100 from a dedicated monitoring security group, and allows all outbound traffic. Monitoring instances use only the monitoring security group.

A security review establishes these requirements:
• Frontend servers may initiate TCP connections only to workers on port 8443.
• Workers may initiate TCP connections only to databases on port 5432.
• Databases must not initiate TCP connections.
• Monitoring instances must retain the ability to initiate TCP connections to all application instances on port 9100.
• Response traffic for permitted connections must work. All other TCP connections initiated by application instances must be blocked, including connections to the internet and corporate networks.

The solution must accommodate instance replacement without maintaining IP address lists. Subnet placement and routing must remain unchanged.

Which security group configuration meets these requirements?

**Correct answer:** Remove the baseline group's self-referencing inbound rule and all its outbound rules, but retain its monitoring inbound rule. Configure frontend egress to reference the worker group on TCP 8443, and worker ingress to reference the frontend group on TCP 8443. Configure worker egress to reference the database group on TCP 5432, and database ingress to reference the worker group on TCP 5432. Remove all other application-tier TCP rules and leave the database group without outbound rules. Allow monitoring-group egress on TCP 9100 to the baseline group.

**Explanation:** The remaining rules permit only the required connection initiators and destinations. Security group references follow group membership as instances are replaced. Removing the baseline group's broad rules is essential because permissions from attached security groups are combined. Security groups are stateful, so responses to allowed connections do not require reverse-direction rules, including database responses and responses to monitoring requests.

### Question 4 (SOA-C03)

**Question:**

A company runs two customers' IPv4-only EC2 workloads in the same subnet of a VPC. Each customer's instances must communicate with one another and with a broker in a separate subnet, but application traffic between the two customers must be blocked. The customers manage their instance security groups, which currently allow all traffic within the VPC.

The workload subnet has a custom network ACL that allows inbound and outbound traffic only to and from the broker subnet's CIDR block. All remaining traffic is denied. Broker connections succeed, but testing shows that the two customers' instances can still communicate directly by using private IPv4 addresses.

A CloudOps engineer must enforce the customer boundary independently of the customer-managed security groups while preserving communication within each customer's workload and with the broker.

Which change meets these requirements?

**Correct answer:** Move one customer's workload into a new, dedicated subnet. Associate that subnet with a custom network ACL that also allows traffic only to and from the broker subnet and denies all remaining traffic.

**Explanation:** Separating the customers into different subnets makes their cross-customer traffic subject to network ACL filtering. Both subnets' ACLs deny that traffic while permitting broker communication. Traffic within each customer's subnet remains available, and customers cannot override the ACL restrictions by changing security groups.

### Question 5 (SCS-C03)

**Question:**

A company connects several production VPCs and a dedicated forensic-analysis VPC to an AWS Transit Gateway. Every production VPC attachment is associated with a Production transit gateway route table. The forensic-analysis VPC attachment is associated with an Investigation transit gateway route table. All attachments propagate their nonoverlapping VPC CIDRs into both tables.

A security engineer identifies a potentially compromised production VPC. During containment, the transit gateway must drop all packets between this VPC and every other production VPC, in both directions. Investigators must still be able to establish connections from the forensic-analysis VPC to the compromised VPC and receive responses. Connectivity among unaffected VPCs must remain unchanged.

VPC routes and resource-level controls already permit the required investigation traffic. There are no alternative connectivity paths, default routes, or forwarding appliances in the attached VPCs. The engineer must implement containment by changing only transit gateway routing.

Which configuration meets these requirements?

**Correct answer:** Create a Quarantine route table containing only a route to the forensic-analysis VPC attachment, and associate the compromised VPC attachment with it. Disable the compromised attachment's propagation into the Production table and verify that no route to its CIDR remains there. Retain its propagation into the Investigation table.

**Explanation:** The association with the Quarantine table restricts packets arriving from the compromised VPC to the forensic-analysis destination. Removing its route from the Production table prevents other production VPCs from sending packets to it. The Investigation table still routes investigator traffic to the compromised VPC, and the Quarantine table provides the response path. Routes used by unaffected VPCs remain unchanged.

### Question 6 (SAA-C03)

**Question:**

A company connects its production and development VPCs through AWS Transit Gateway. Security policy requires all traffic between these environments to pass through stateful firewall appliances in a dedicated inspection VPC. The inspection VPC has a transit gateway attachment subnet and a Gateway Load Balancer endpoint in each of two Availability Zones.

Transit gateway route tables direct traffic from the workload VPCs to the inspection VPC and then to the destination workload VPC. No direct route bypasses inspection. The firewall configuration and inspection VPC subnet routes have been validated. However, some connections fail because the transit gateway sends forward and return traffic through different Availability Zones in the inspection VPC.

Which change will meet the inspection requirement and reliably support these stateful connections?

**Correct answer:** Enable appliance mode support on the inspection VPC's transit gateway attachment.

**Explanation:** Appliance mode makes the transit gateway maintain Availability Zone affinity for a flow through the inspection VPC attachment. This keeps forward and return traffic on the same inspection path, supporting stateful appliances while retaining the required segmentation.

### Question 7 (DOP-C02)

**Question:**

A company uses an infrastructure-as-code pipeline to attach production, development, and shared-services VPCs to one AWS Transit Gateway. Each VPC has a nonoverlapping IPv4 CIDR range. The VPC subnet route tables already direct traffic for the other VPC CIDRs to the transit gateway. There are no other connections between the VPCs.

Production and development workloads must each communicate bidirectionally with shared services. However, production and development must have no routing path to each other, even if workload security groups are accidentally configured to allow the traffic. The shared-services VPC does not contain a firewall or routing appliance.

The DevOps engineer must automate this boundary for existing attachments and future attachments classified as production, development, or shared services. Default transit gateway route table association and propagation are disabled.

Which transit gateway configuration should the pipeline implement?

**Correct answer:** Create a workload route table and a shared-services route table. Associate production and development attachments with the workload table, and propagate only shared-services routes into it. Associate shared-services attachments with the shared-services table, and propagate production and development routes into that table. Apply these rules automatically according to each new attachment's classification.

**Explanation:** Traffic entering from either workload environment can reach shared services but has no route to the other environment. Traffic entering from shared services has routes back to both environments. Classifying new attachments and applying the same association and propagation rules preserves the boundary automatically.

### Question 8 (ANS-C01)

**Question:**

A company connects workload VPCs in several AWS accounts to a central transit gateway. Each workload VPC has a unique /16 CIDR within 10.64.0.0/12. A shared-services VPC uses 10.80.0.0/16. A separate egress VPC provides centralized internet access.

All workload attachments are associated with a transit gateway route table named Spoke. Every workload attachment and the shared-services attachment propagate their VPC CIDRs into Spoke. Spoke also contains a static 0.0.0.0/0 route to the egress attachment. Separate transit gateway route tables already provide the required return paths from the shared-services and egress VPCs. VPC routes and security controls permit connectivity.

The company must prevent communication between workload VPCs while preserving access from every workload VPC to shared services and the public internet. Packets destined for another workload VPC must be discarded by the transit gateway when they first arrive from a workload attachment, without being forwarded to the egress VPC. The network team reserves 10.64.0.0/10 for private networks that must not be reached through the default egress route, except for explicitly permitted, more-specific destinations.

Which change to Spoke meets these requirements?

**Correct answer:** Disable propagation from every workload attachment into Spoke and add a static blackhole route for 10.64.0.0/10. Retain shared-services propagation and the default route to the egress attachment.

**Explanation:** Disabling workload propagation removes the workload /16 routes from Spoke. The /10 blackhole then discards workload-to-workload traffic instead of allowing it to follow the default route. The shared-services /16 route is more specific than the blackhole and remains reachable. Public internet destinations continue to use the default route.

## Practice Network Segmentation in a full exam

Questions on Network Segmentation in each practice exam, with every answer option and an explanation for each:

- [SAA-C03](https://blowstack.com/practice-exams/aws-certified-solutions-architect-associate): 12 questions
- [ANS-C01](https://blowstack.com/practice-exams/aws-certified-advanced-networking-specialty): 11 questions
- [SCS-C03](https://blowstack.com/practice-exams/aws-certified-security-specialty): 8 questions
- [MLA-C02 (beta)](https://blowstack.com/practice-exams/aws-certified-machine-learning-engineer-associate): 3 questions
- [SOA-C03](https://blowstack.com/practice-exams/aws-certified-cloudops-engineer-associate): 3 questions
- [DOP-C02](https://blowstack.com/practice-exams/aws-certified-devops-engineer-professional): 3 questions
- [CLF-C02](https://blowstack.com/practice-exams/aws-certified-cloud-practitioner): 2 questions
- [SAP-C03](https://blowstack.com/practice-exams/aws-certified-solutions-architect-professional): 2 questions

## What is Network Segmentation

Network segmentation creates boundaries between groups of systems so that connectivity follows security requirements instead of being universally available. Typical boundaries separate production from development, internet-facing components from internal services, and application servers from sensitive databases. The objective is to reduce the paths an attacker can use and limit the blast radius of a compromised workload. Start by identifying required communication: which source must reach which destination, on what protocol and port, and for what operational purpose.

Cloud networks implement these boundaries using virtual networks, subnets, routing, and traffic filtering. In AWS, an Amazon VPC provides a logically isolated network, and each subnet occupies one Availability Zone. Route tables determine where subnet traffic is directed. An internet gateway supports direct internet connectivity, while a NAT gateway can provide outbound IPv4 internet access for workloads without exposing them to unsolicited inbound connections through that gateway. NAT is not a substitute for traffic filtering or application-aware egress inspection.

A public subnet has a route to an internet gateway. That designation does not automatically expose every resource: direct IPv4 internet communication also requires suitable public addressing and permissive security controls. A private subnet lacks a direct route to an internet gateway but may have outbound access through NAT. An isolated subnet has no route for internet access, including through NAT, although it can retain explicitly permitted private connectivity. Private service endpoints can provide access to supported cloud services without requiring an internet route.

Subnets alone do not enforce application isolation. Within an AWS VPC, route tables normally include local routing between its address ranges. Security groups provide stateful filtering on associated network interfaces, while network ACLs provide stateless subnet-level filtering. Security groups contain allow rules; network ACLs support allow and deny rules and require explicit consideration of return traffic. Network ACLs do not filter traffic between resources in the same subnet, so workload-level controls remain important.

North/south traffic crosses a defined environment boundary, such as traffic between the internet and an application. East/west traffic moves between workloads or internal segments. Protecting an internet entry point does not prevent lateral movement between application servers and databases. For a three-tier application, permit HTTPS to the public load balancer, permit application traffic only from the load balancer to the application tier, and permit database connections only from the application tier. Where supported, security group references can express these relationships without relying on changing instance IP addresses.

Stronger boundaries may use separate virtual networks or accounts, with explicitly controlled connectivity through a transit network or inspection firewall. Central inspection only works when routing actually sends the relevant traffic through the firewall; stateful inspection also requires consistent forward and return paths. Additional boundaries increase route management, policy maintenance, troubleshooting effort, and potentially cost and latency. Validate both permitted and prohibited paths using connectivity analysis, flow logs, and controlled tests. When troubleshooting, inspect routes, security groups, network ACLs, firewall policy, addressing, and return paths rather than assuming that a private subnet is inherently isolated.

## Learn Network Segmentation in lessons

10 short lessons on Network Segmentation in 7 certification courses.

- CLF-C02: Network Segmentation: workload boundaries in a VPC (Premium), https://app.blowstack.com/lessons/network-segmentation-workload-boundaries-in-a-vpc-clf-c02
- MLA-C02 (beta): Network Isolation vs. Network Segmentation (Premium), https://app.blowstack.com/lessons/network-isolation-vs-network-segmentation-mla-c02
- SAP-C03: Network Segmentation vs. Subnetting (Premium), https://app.blowstack.com/lessons/network-segmentation-vs-subnetting-sap-c03
- SOA-C03: Network Segmentation: enforcing workload boundaries (Premium), https://app.blowstack.com/lessons/network-segmentation-enforcing-workload-boundaries-soa-c03
- SCS-C03: Network Segmentation: enforcing trust boundaries (Premium), https://app.blowstack.com/lessons/network-segmentation-enforcing-trust-boundaries-scs-c03
- SAA-C03: Network Segmentation: enforcing workload trust boundaries (Premium), https://app.blowstack.com/lessons/network-segmentation-enforcing-workload-trust-boundaries-saa-c03
