What is a software-defined perimeter (SDP)?
A software-defined perimeter (SDP) is a way to hide Internet-connected infrastructure (servers, routers, etc.) so that external parties and attackers cannot see it, whether it is hosted on-premises or in the cloud. The goal of the SDP approach is to base the network perimeter on software instead of hardware. A company that uses an SDP is essentially draping a cloak of invisibility over their servers and other infrastructure so that no one can see it from the outside; however, authorized users can still access the infrastructure.
A software-defined perimeter forms a virtual boundary around company assets at the network layer, not the application layer. This separates it from other access-based controls that restrict user privileges but allow wide network access. Another key difference is that an SDP authenticates devices as well as user identity. The Cloud Security Alliance first developed the SDP concept.
How does an SDP work?
With an SDP, it should not be technologically possible to connect with a server unless authorized to do so. SDPs allow access to users only after 1) verifying user identity, and 2) assessing the state of the device.
Once the user and device are authenticated, the SDP sets up an individual network connection between that device and the server it is trying to access. An authenticated user is not logged into a larger network, but rather is given their own network connection that no one else can access and that only includes the services that the user has approved access to.
Imagine a web server that is connected to the Internet but does not open connections with anything. It does not accept requests or send responses; it has no open ports and no network access even though it is plugged into the Internet (somewhat like a toaster or a lamp that is plugged into a wall outlet but turned off so that electricity does not flow through it). This is the default state for servers inside a software-defined perimeter.
Another way to think of SDPs is to imagine a front door that is always kept locked. No one can come through the door, or even look inside, until the person on the other side of the door verifies who the visitor is and what they are doing. Once the visitor is allowed inside, the person in the house locks the door again.
How does a user gain access over an SDP?
User identity verification: User identity is typically verified via a third-party identity provider (IdP). SDPs may integrate with an SSO solution as well. User authentication can involve a simple username and password combination, but it is more secure to use multi-factor authentication with some sort of hardware token.
Device verification: This involves checking to make sure the user's device is running up-to-date software, checking for malware infections, and performing other security inspections. Theoretically, an SDP could even create a blacklist of disallowed devices and check to make sure the device is not on the blacklist.
SDP controller approval: The SDP "controller" is the logical component of the SDP that is responsible for determining which devices and servers should be allowed to connect. Once the user and device are authenticated, the controller passes along approval of the user and device to the SDP gateway. The SDP gateway is where access is actually allowed or denied.
Secure network connection established: The SDP gateway opens the virtual "gate" to allow the user through. It establishes a secure network connection with the user device on one side of the gateway, and on the other side it establishes a network connection with the services that the user has access to. No other users or servers share this connection. These secure network connections typically involve the use of mutual TLS, and may use a VPN.
User access: The user is able to access previously hidden network resources and can continue using their device like normal. The user operates within an encrypted network to which only they and the services they access belong.
What is mutual TLS?
TLS, or Transport Layer Security, is an encryption protocol that verifies that a server is legitimate, meaning it is who it claims to be. In mutual TLS, the verification process takes place on both sides: the client is verified as well as the server. Learn more about mutual TLS.
What is a VPN?
A VPN, or virtual private network, is an encrypted network that runs over an unencrypted network. It creates encrypted connections between devices and servers so that it is as if they are on their own private network.
Traditionally, VPNs have been used to secure and manage access to company infrastructure. In some cases, an SDP can replace a VPN.
SDP vs. VPN: What are the differences?
SDPs may incorporate VPNs into their architecture to create secure network connections between user devices and the servers they need to access. However, SDPs are very different from VPNs. In some ways, they are more secure: while VPNs enable all connected users to access the entire network, SDPs do not share network connections. SDPs may also be easier to manage than VPNs, especially if internal users need multiple levels of access.
Managing several different levels of network access using VPNs involves deploying multiple VPNs. Suppose Bob works at Acme Inc. in accounting, Carol works in sales, and Alice works in engineering. Managing their access at the network level involves setting up three VPNs: 1) an accounting VPN to provide access to the accounting database, 2) a sales VPN for the customer database, and 3) an engineering VPN for the codebase. Not only is this difficult for IT to manage, it is also less secure: anyone who gains access to the accounting VPN, for instance, can now access the financial information of Acme Inc. If Bob accidentally gives his login credentials to Carol, then security is compromised — and IT may not even be aware of it.
Now imagine a fourth person: David, the CFO of Acme Inc. David needs access to both the accounting database and the customer database. Should David have to log into two separate VPNs in order to do his work? Or should Acme's IT department set up a new VPN that provides access to both accounting and the customer database? Both options are unwieldy to manage, and the second option introduces a high degree of risk: an attacker who compromises this new VPN with broad access across two databases is able to do twice as much damage as before.
SDPs, on the other hand, are much more granular. There is no overall VPN that everyone who accesses the same resources logs into; instead, a separate network connection is established for each user. It is almost as if everyone has their own private VPN. In addition, SDPs verify devices as well as users, making it much more difficult for an attacker to breach the system with stolen credentials alone.
A couple other key features separate SDPs from VPNs: SDPs are location- and infrastructure-agnostic. Because they are based on software rather than hardware, SDPs can be deployed anywhere to protect on-premises infrastructure, cloud infrastructure, or both. SDPs also easily integrate with multi-cloud and hybrid cloud deployments. And finally, SDPs can connect users in any location; they do not need to be within a company's physical network perimeter.
How do SDPs relate to zero trust security?
As the name implies, there is no trust in zero trust security; no user, device, or network is considered trustworthy by default. Zero trust security is a security model that requires strict identity verification for every person and device trying to access internal resources, no matter whether they are sitting inside or outside the network perimeter (or, the software-defined perimeter).
An SDP is one way to implement zero trust security. Users and devices must both be verified before they can connect, and they have only the minimum network access they need. No device, not even a CEO's laptop, can form a network connection with a resource it is not authorized to use.