Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Hybrid cloud security protects estates that span on-premises systems and public cloud, where the hardest problems are the identity bridge and the connectivity joining them.

What Is Hybrid Cloud Security?
Hybrid cloud security covers estates where workloads, data, and identity are split between on-premises infrastructure and one or more public clouds. Most large UK organisations are in this position and will remain there: a mainframe or line-of-business application that will not move, a data centre lease with years to run, latency or regulatory reasons to keep certain processing local. The security problem is not either environment individually. It is the seams.
Hybrid estates almost always share identity, usually by synchronising Active Directory into Microsoft Entra ID or by federating through AD FS. That bridge is convenient, necessary, and the single most valuable target in the environment, because it connects two trust domains that were designed independently.
The SolarWinds intrusion disclosed in December 2020 showed what that looks like in practice. Having compromised on-premises infrastructure, the actor stole the AD FS token-signing certificate and forged SAML assertions—the technique known as Golden SAML—to authenticate to cloud services as any user, without ever touching the cloud authentication path. No password, no second factor, no failed login to detect.
Practical defences follow from the mechanism. Protect token-signing material in a hardware security module and rotate it. Monitor for assertions with unusual lifetimes or issuers. Prefer cloud-native authentication, with pass-through or cloud-managed credentials, over federation where you can. Treat domain controllers and federation servers as tier-zero assets with the same controls you would apply to a cloud management account, and read our entry on identity provider federation risks alongside this one.
Site-to-site VPN, AWS Direct Connect, and Azure ExpressRoute exist to make the cloud feel like an extension of the corporate network. They succeed, and that is the problem: a compromised laptop on the corporate LAN often has an unobstructed path to cloud workloads, and a compromised cloud workload frequently has a path back.
Treat the link as a boundary rather than a merge. Apply firewalling and inspection in both directions, define explicitly which subnets and services may traverse it, avoid flat routing between the corporate network and cloud virtual networks, and apply the design thinking in cloud network segmentation to both sides. Where the only requirement is access to a handful of services, private endpoints or an application proxy beat a full network bridge.
Two environments tend to produce two of everything: two patching regimes, two change processes, two vulnerability tools, two sets of logs with different schemas, and two teams who each assume the other is covering the overlap. Attackers find the gap because it is usually the same gap—an asset that belongs to neither inventory.
Aim for one asset register covering both sides, one risk acceptance process, one set of baseline standards expressed per platform, and one place where alerts land. Perfect tooling uniformity is unrealistic and not the point; agreed ownership and a single correlation layer are. Microsoft Defender for Cloud, Google Anthos, AWS Systems Manager with hybrid activations, and Azure Arc all extend cloud management to on-premises machines, which is often easier than extending on-premises tooling into the cloud.
Classify before you decide placement, not afterwards. Know which datasets may cross the link, encrypt them in transit with keys you control, and keep a clear record of where each sensitive dataset lives for the reasons set out in data residency and data sovereignty. Backup deserves specific attention: a hybrid estate frequently backs cloud workloads to on-premises storage or vice versa, and the resulting dependency is only visible during a recovery, when the link is down.
At Privia Security, we test hybrid attack paths end to end through our infrastructure security testing services.
Golden SAML is an attack in which an adversary steals the token-signing certificate from a federation server such as AD FS and forges authentication assertions for any user. Because the token is signed by a trusted issuer, the cloud service accepts it without a password or second factor, and no failed login appears. It is the clearest reason to treat federation infrastructure as a tier-zero asset.
It is a trust relationship, and an unfiltered one is a risk. A flat route between the corporate network and cloud virtual networks means a compromised endpoint reaches cloud workloads directly, and a compromised cloud workload can return. Apply firewalling and inspection in both directions, restrict which subnets and ports may traverse the link, and consider private endpoints or an application proxy where full network connectivity is not required.
Where practical, yes, but consistency of ownership matters more than consistency of product. One asset inventory, one risk process, one set of baselines expressed per platform, and one place where alerts are correlated will do more than forcing a single agent onto every machine. Cloud management planes that extend on-premises, such as Azure Arc or Systems Manager, often make this easier than the reverse.
At the boundary. The cloud team cannot see on-premises logs, the infrastructure team has no cloud access, the two log sets use different schemas and time sources, and nobody has agreed who leads when an incident crosses over. Rehearse a scenario that starts on one side and moves to the other, and fix the access, correlation, and decision-making gaps that the exercise exposes.
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.