Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Cloud network segmentation divides a cloud estate into isolated zones using accounts, virtual networks, security groups, and private endpoints, so one compromise cannot spread.

What Is Cloud Network Segmentation?
Cloud network segmentation is the division of a cloud estate into zones that cannot freely reach one another. The purpose is containment: when one workload is compromised—and eventually one will be—the attacker should find a wall rather than an open route to the database, the build system, and the identity platform. The mechanisms differ from a traditional data centre, but the goal is the one described in our entry on network segmentation.
Segmentation in the cloud happens at several levels at once, and the strongest ones are not network controls at all.
Almost every flat estate was built by well-meaning people under deadline. A single large VPC is simpler to set up. Address space was allocated before anyone knew how many environments there would be. A security group with 0.0.0.0/0 on port 443 was a temporary fix during an incident. Peering was added to solve one integration and left in place. Each decision was defensible; the aggregate is an environment where a compromised test server can reach production.
The same pattern appears in hybrid estates, where a site-to-site connection between the corporate network and a cloud virtual network quietly becomes a two-way bridge. That is one of the routes discussed in hybrid cloud security.
Inbound rules get attention because they map to the obvious threat. Outbound rules matter at least as much, because command and control, data exfiltration, and malware download all travel outbound. Default allow-all egress means a compromised container can reach any host on the internet.
Practical egress control means no internet gateway on private subnets, a controlled NAT or proxy path for the traffic that genuinely needs the internet, an allowlist of destinations for sensitive workloads, and private endpoints for provider services so that traffic to storage or a key vault never leaves the provider network. Logging egress is valuable even where blocking is impractical.
Network position is a weaker control in the cloud than it was on-premises, because most access happens through authenticated APIs that ignore your subnets entirely. A role with read access to a storage bucket does not care which network it is called from unless you add a condition saying so—and adding those conditions, tying API access to a specific VPC endpoint or address range, is one of the highest-value controls available. Segmentation and identity reinforce each other; neither is sufficient alone.
At Privia Security, we validate segmentation boundaries with hands-on testing through our network security services.
A security group is stateful and applies to an instance or network interface, so return traffic for an allowed connection is permitted automatically and you only write allow rules. A network ACL is stateless and applies to a whole subnet, evaluates rules in order, supports explicit denies, and requires you to allow return traffic separately. Most designs use security groups primarily and ACLs for coarse blocks.
Generally yes. Separate accounts remove the implicit identity relationships that make lateral movement easy, give you independent guardrails and quotas, and limit the damage a compromised administrator can do. Separate VPCs within one account still share an identity boundary, so a role with wide permissions reaches both. Use account separation for trust boundaries and VPCs for network structure within them.
Yes, for anything handling sensitive data. Unrestricted outbound access lets a compromised workload contact command and control infrastructure, download tooling, and exfiltrate data without touching a single inbound rule. Private endpoints for provider services, a controlled proxy or NAT path, and destination allowlists for regulated workloads cover most of the risk. Log outbound traffic even where you cannot yet block it.
Test it from inside. Launch a host in each zone and attempt to reach the services you believe are unreachable, including provider APIs, management endpoints, and the metadata service. Review effective rules with provider tools such as AWS Network Access Analyzer or Azure network watcher. PCI DSS requires this at defined intervals, and it is worth doing regardless of whether a standard obliges you.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.