Skip to main content
  • blog
  • CMMC Enclave Architecture: A Practical Guide to Building, Configuring, and Managing a Compliant Enclave

CMMC Enclave Architecture: A Practical Guide to Building, Configuring, and Managing a Compliant Enclave

  • June 16, 2026
Author

Emily Bonnie

Senior Content Marketing Manager

If you've decided that a CUI enclave is the right path to CMMC Level 2 certification, you've already made one of the most important strategic decisions in your compliance journey. Scoping your environment down to a defined, secured boundary is one of the most effective ways to reduce cost, complexity, and certification timelines.

But deciding to build an enclave and actually building one are two very different things. An enclave that isn't properly architected, configured, and maintained can create a false sense of security and raise issues during a C3PAO assessment.

If you're still deciding whether an enclave is right for your organization, start with our guide to CUI enclaves, then come back here. This guide gives a clear, practical picture of what a defensible CMMC enclave looks like: what goes into it, how it needs to be configured, and how you keep it compliant over time.

Step 1: Defining scope

The most expensive mistake in CMMC compliance isn't choosing the wrong cloud platform or missing a configuration setting. It's building before you've clearly defined what needs to be protected and where it lives.

An enclave makes the most sense when a relatively small percentage of your workforce handles CUI and federal contracts represent a meaningful but not dominant portion of your revenue. If most of your team handles CUI regularly and the majority of your revenue comes from federal work, applying CMMC controls enterprise-wide may actually be simpler than maintaining two separate environments. 

Incomplete scoping is one of the most common findings during CMMC Level 2 assessments. If a C3PAO discovers that CUI exists outside your defined enclave boundary, your scope can expand unexpectedly mid-assessment, pulling in additional systems, endpoints, and users that now have to meet the full set of 110 NIST SP 800-171 controls.

Before you provision a single resource, you need to map your CUI.

Controlled Unclassified Information (CUI) is sensitive government-related data that isn't classified but still requires protection under regulations like DFARS 252.204-7012. This includes engineering specifications, procurement records, export-controlled technical data, and more. If your contracts involve DoD work, you likely handle CUI. The question is where it lives and how it moves through your environment.

Map your CUI flows

CUI flow mapping means identifying every person, system, and process that touches CUI: where it enters your environment, where it lives, how it moves, and where it exits. This includes:

  • File servers, databases, and cloud storage where CUI is stored
  • Email platforms and communication tools used to transmit CUI
  • Applications used to generate, view, edit, or share CUI (CAD tools, document management systems, project management platforms)
  • Users who handle CUI in any capacity
  • Any integrations or automated workflows that move CUI between systems
  • Printers, removable media, and physical documents containing CUI

It’s important to note that CUI mapping isn’t a one-time exercise. It becomes the foundation of your System Security Plan and your ongoing compliance program. If your documentation doesn't match how your environment actually operates, assessors will find the gaps.

Categorize your assets

Once you've mapped your CUI flows, you'll categorize every asset in your environment. The CMMC scoping guide defines four asset categories, and this categorization determines what level of scrutiny each asset receives during an assessment:

  1. CUI Assets are systems that store, process, or transmit CUI. Every CUI asset must meet all 110 NIST SP 800-171 controls. This is what lives inside your enclave.
  2. Security Protection Assets (SPAs) provide security functions for your enclave (firewalls, SIEM tools, identity management systems). They don't handle CUI directly, but they protect the systems that do. SPAs are in scope for CMMC and must be documented in your SSP.
  3. Contractor Risk Managed Assets (CRMAs) don't process CUI but could indirectly affect the security of your enclave if compromised. These are managed through risk-based controls rather than the full NIST 800-171 control set.
  4. Out-of-Scope Assets (OSAs) have no connection to your CUI environment. The goal of your enclave strategy is to maximize the number of systems that fall into this category and outside of your assessment boundary.

The cleaner this categorization, the smaller your assessment footprint.

Recommended reading

An Expert’s Guide to CMMC Level 2 Scoping and Asset Categories

Cloud-only vs. hybrid environments

The shape of your enclave depends heavily on whether you have CUI that lives only in digital systems or whether you also have physical CUI such as manufacturing specifications on a shop floor, controlled technical documents on secure printers, or CNC machines running export-controlled programs.

Cloud-only enclaves keep everything in a FedRAMP-authorized cloud environment, and users access CUI through virtual desktops or managed devices. Physical endpoints stay out of scope, there’s no on-premises infrastructure to secure and document, and the cloud provider’s FedRAMP SSP covers a large portion of shared controls. This is the fastest path to compliance and the simplest to maintain, and it's the right choice for most knowledge workers and office-based teams.

Hybrid (bridged) enclaves connect a cloud enclave to on-premises systems via a site-to-site VPN. This brings your physical site into CMMC scope, which means physical access controls, network segmentation at the facility level, and a larger documentation burden. It's the right choice when you must handle physical CUI that can't be moved to the cloud. For example, you operate CNC machines, 3D printers, or other manufacturing equipment that processes export-controlled technical data, you have a secure print environment for CUI documents, or you run lab equipment or test systems with CUI data that can't be moved to the cloud. 

In a hybrid model, you'd typically designate a single secure room or facility area as the physical enclave boundary (rather than the entire building). Intune or your MDM solution extends from the cloud enclave to manage physical workstations within that boundary. Physical access controls, visitor logs, and clean desk policies become part of your compliance program.

The tradeoff is real: every piece of physical infrastructure within the boundary must meet CMMC requirements and be documented in your SSP. 

Step 2: Choosing your cloud platform

Once you've mapped your CUI and decided on your enclave model, you need a cloud foundation. For CMMC Level 2 that means a FedRAMP-authorized environment that meets the security requirements for handling CUI. Two platforms are well-suited to this are:

Microsoft 365 GCC High + Azure Government

Microsoft 365 GCC High is the most widely deployed platform for CMMC-compliant enclaves. It's specifically designed for organizations subject to DFARS 252.204-7012, ITAR, EAR, and CMMC, and it runs on infrastructure that's physically separate from Microsoft's commercial cloud.

With GCC High:

  • Data is stored exclusively in US-based data centers staffed by US persons
  • The environment is FedRAMP High authorized
  • It inherits Microsoft's FedRAMP High System Security Plan, which covers a significant portion of your shared responsibility
  • It includes Exchange Online, SharePoint, OneDrive, and Teams within the GCC High boundary
  • It's the only Microsoft environment that supports ITAR/EAR data

A complete GCC High enclave stack typically includes:

  • Microsoft 365 GCC High for email, collaboration, and document storage
  • Azure Government subscription (separate from commercial Azure) for infrastructure, compute, and services
  • Azure Virtual Desktop (AVD) for virtual workstations
  • Azure Premium Firewall for network perimeter protection
  • Microsoft Sentinel for SIEM and log management 
  • Microsoft Defender suite for endpoint, identity, and cloud security
  • Entra ID (Azure AD) for identity management, conditional access, and MFA
  • Microsoft Intune for device management and compliance policies
  • Microsoft Purview for data loss prevention, sensitivity labels, and data classification

One critical note on Microsoft licensing: GCC High runs on E3 or E5 licensing, which is more expensive than commercial M365 plans. For organizations evaluating cost, GCC High licensing is typically a significant line item. E5 licensing adds advanced security capabilities (Defender, Sentinel, Purview) that satisfy many CMMC controls out of the box, which often makes it cost-effective when you factor in what you'd otherwise spend on third-party tools.

If you're currently on GCC (not GCC High) and you have any ITAR data or think you might in the future, start on GCC High. Migrating from GCC to GCC High after the fact is a significant undertaking.

Recommended reading

GCC High Alternatives for CMMC: Cloud Options Compared

Google Workspace for Government

Google Workspace for Government (GWS) is a strong option for contractors already operating in Google-native environments or on Google Cloud Platform. It's FedRAMP High authorized, meaning it already meets the security baseline required for handling CUI.

GWS offers:

  • Gmail, Drive, Docs, Meet, and the broader Workspace suite within a FedRAMP High boundary
  • US-only data residency
  • Strong identity management via Google Identity and support for third-party MFA
  • Native integration with Google Cloud Platform for organizations running infrastructure on GCP

For organizations that have built workflows around Google tools and want to minimize migration overhead, GWS can serve as the cloud foundation for a CMMC enclave. You'll still need to configure the additional controls required by NIST SP 800-171, including SIEM, endpoint management, DLP, and boundary protection, but the FedRAMP High authorization provides a solid compliance starting point.

GWS doesn't have a native virtual desktop offering equivalent to Azure Virtual Desktop. Organizations choosing GWS will typically rely on an MDM solution for endpoint management or a third-party VDI tool for virtual workstations.

Recommended reading

cmmc-hero

Does Google Workspace Meet CMMC Requirements?

Step 3: Building core architecture

Regardless of which cloud platform you choose or whether your enclave is cloud-only or hybrid, there are architectural components that map directly to NIST 800-171 control requirements that every CMMC-compliant enclave must include.

Identity and Access Management

In a cloud enclave, there is no locked server room or on-premises network to hide behind. The only thing standing between an attacker and your CUI is your identity layer. That's why CMMC dedicates significant requirements to it: if identity controls fail, every other security measure can be bypassed. Every access decision in your enclave flows through identity, which means it also has to be where your tightest controls live.

  • Separate enclave identities from commercial identities. Users who need access to both your commercial environment and your CUI enclave need separate accounts for each. Sharing identities across environments is a boundary violation: if a commercial account is compromised, it can't be a path into your enclave.
  • Enforce MFA everywhere, without exceptions. CMMC requires multi-factor authentication for all users accessing CUI systems. Configure MFA at the identity provider level so it can't be bypassed at the application layer. Every session, every time.
  • Implement Role-Based Access Control (RBAC). Users should have access only to the CUI assets they need to do their jobs. Least privilege is both a NIST requirement and a practical defense against spillage. Define roles before you provision users, and audit those roles regularly.
  • Configure conditional access policies. Conditional access allows you to enforce additional requirements based on context, such as blocking access from non-compliant devices, requiring step-up authentication for sensitive resources, or restricting access to US-only IP ranges. This is one of the most powerful tools in the Entra ID/Azure AD toolkit.
  • Plan for "swivel seat" users. Some employees legitimately need access to both your commercial Microsoft tenant and your GCC High enclave. Cross-tenant access is one-way by design: GCC High accounts can read commercial resources (SharePoint, OneDrive, federated Teams chat), but commercial accounts cannot access GCC High. Dual licensing is required for users who work in both environments. Document this access pattern in your System Security Plan (SSP).

Network Segmentation

Your enclave boundary needs to be technically enforced, not just documented. A policy that says "CUI stays in the enclave" means nothing if there's no technical mechanism preventing it from crossing the boundary.

  • Deploy a next-generation firewall. In Azure Government, Azure Premium Firewall provides stateful packet inspection, application-layer filtering, and threat intelligence integration. It sits at the perimeter of your enclave and controls all traffic in and out.
  • Configure Network Security Groups (NSGs). NSGs provide subnet-level access control within your Azure environment, restricting which resources can communicate with which. Apply NSGs to every subnet in your CUI environment.
  • Control external internet access. CUI systems should not have unrestricted access to the public internet. Route external traffic through a secure proxy or restrict it to known-good destinations. Any data transfer mechanism that allows CUI to leave the enclave uncontrolled is a boundary failure.
  • For hybrid enclaves, configure a site-to-site VPN. The VPN tunnel between your cloud enclave and on-premises environment needs to meet FIPS 140-2 encryption requirements. Document the VPN configuration in your SSP, including the traffic that's authorized to traverse it.

Endpoint Controls

Every device that can access CUI is a potential path for CUI to escape the enclave. Your endpoint strategy determines whether physical devices enter your CMMC scope or stay out of it entirely. CUI should only ever be accessible through a controlled, documented, hardened endpoint.

Data Protection

CUI has to stay protected wherever it lives and however it moves: in storage, in transit, and at the application layer. These controls address what happens to the data itself, independent of who has access to it.

  • Encrypt data at rest and in transit. In Azure Government, encryption is largely handled at the platform level, but you need to verify key management practices, specifically that you control your own encryption keys via Azure Key Vault rather than relying on Microsoft-managed keys.
  • Deploy Data Loss Prevention (DLP). DLP is how you enforce the access control requirement that CUI only moves to authorized destinations. Microsoft Purview (included in GCC High E5) lets you create policies based on how data is classified. For example, you can apply a "CUI" sensitivity label to files and emails, then build DLP rules that automatically block any labeled content from being sent to external email addresses, shared via a public link, or downloaded to a device that isn't enrolled in your MDM. The rules apply at the application layer, so they work even if a user tries to route around your network controls.
  • Control removable media. USB devices are a common CUI exfiltration vector. Your enclave policy should block USB storage by default and allow exceptions only through documented, managed processes. In Azure Virtual Desktop environments, disable USB redirection entirely.

Logging and Monitoring

NIST SP 800-171's Audit and Accountability family requires that every action taken by every user in your enclave is logged, traceable back to that user, and protected from tampering. This is one of the more demanding control families in practice and one of the most important, because it's how you detect a breach, investigate an incident, and prove to an assessor that your controls are actually working.

  • Centralize all logs. Every enclave system, including identity, network, endpoints, and applications, should send logs to a central Security Information and Event Management (SIEM) that collects, correlates, and analyzes security logs from across your environment. In GCC High, Microsoft Sentinel is the natural choice, and Microsoft-hosted service logs (Azure AD, Defender, Office 365) feed into Sentinel with no ingestion costs.
  • Log according to NIST 800-171 3.3 requirements. At minimum, you need to log: user logins and logouts, failed authentication attempts, privileged account activity, file access events for CUI data, administrative changes to security configurations, and network boundary crossings. 
  • Protect logs from tampering. Audit logs must be protected against unauthorized modification or deletion. In Sentinel, configure log retention according to your data retention policy (minimum 90 days accessible, longer for archival) and restrict who can modify or delete log data.
  • Set up real-time alerting. Configure alerts for high-priority events including failed MFA attempts, impossible travel logins, bulk file downloads, changes to privileged accounts, and firewall rule modifications. Your incident response plan should define how you respond to each alert type.

Backup and Recovery

CMMC Level 2 requires regularly performing and testing data backups and protecting the confidentiality of backup CUI at storage locations. Level 3 adds requirements for complete and comprehensive resilient backups.

According to current documentation, Microsoft's native retention has gaps that matter for CMMC compliance:

Service Native Retention
Deleted accounts 30 days
OneDrive 30 days accessible by default after account deletion (configurable); moves to recycle bin for 93 additional days, PowerShell restore only
SharePoint Versioning and limited site collection restore (14-day window, support ticket required); native Microsoft 365 Backup not available in GCC High
Azure Site Recovery equivalent for O365 N/A

For most CMMC programs, native O365 retention won't be sufficient. SharePoint's built-in restore options are limited to item-level version history and a 14-day site collection restore window that requires a Microsoft support ticket, and the native Microsoft 365 Backup service available in commercial cloud isn't offered in GCC High. 

You'll need a third-party backup solution that's FedRAMP authorized, hosts data in US government data centers, and supports your required recovery point objectives (RPOs) and recovery time objectives (RTOs). Verify that any backup vendor you consider has FedRAMP Moderate authorization at minimum and FedRAMP High for ITAR data.

Step 4: Define your endpoint protection strategy

How your users access CUI determines whether their physical devices end up in your CMMC assessment scope. This is one of the highest-leverage decisions in your enclave architecture, and it can dramatically reduce the number of assets you have to harden, document, and maintain.

You have two primary options: virtual desktops and mobile device management. Understanding what each one does, what it doesn't do, and when to use each is essential.

Virtual Desktops (VDI)

Azure Virtual Desktop (AVD) is Microsoft's hosted virtual desktop service, running inside Azure Government. Instead of working on their local laptop, a user opens a Remote Desktop client  and connects to a full Windows desktop that lives entirely in the cloud. The session runs on the virtual machine and the user's local laptop is just a display.

For endpoints to stay out of scope, your Azure Virtual Desktop configuration must enforce the following settings:

  • Disable clipboard redirection: Users cannot copy content from the virtual desktop to their local machine
  • Disable drive redirection: Users cannot save files from the virtual desktop to local disk
  • Disable printer redirection: Unless printing to a FIPS 800-171-compliant print system
  • Disable USB redirection: No USB devices can be passed through to the virtual session
  • Disable folder redirection: No local folder syncing
  • Enable screen capture protection: Prevents screenshots of the virtual desktop session. 
  • Require MFA on every session: Authentication can't be bypassed between sessions

One important caveat on client support: based on current Microsoft documentation, screen capture protection works automatically on Windows and macOS using the Windows App or Remote Desktop client. iOS/iPadOS and Android devices are also supported but require additional configuration, specifically, an Intune MAM app protection policy set to block screen capture (minimum Windows App version 11.2.4 for iOS, 11.0.0.94 for Android). Web browsers cannot connect at all when screen capture protection is enabled.

If your CUI users access Azure Virtual Desktop from mobile devices without the required MAM configuration in place, their devices cannot be used to keep physical endpoints out of scope. Verify your Intune configuration carefully before treating mobile users as out of scope.

Host pool configurations:

When you deploy virtual desktops in AVD, you configure what's called a host pool: a group of virtual machines that users connect to when they log in. You'll configure either pooled (N:1, load-balanced, where multiple users share virtual machines) or personal (1:1, where each user gets a dedicated virtual machine) pools. For most office productivity use cases, pooled is appropriate. For GPU-intensive work like CAD or simulation, you'll need personal pools with NV-series GPU-optimized virtual machines.

Keep one base image per host pool to ensure consistent security configurations: different images on different hosts means users could get different security settings on different logins. The number of users per virtual machine is set at pool creation and cannot be changed later, so size carefully.

User profile management:

One practical challenge with pooled virtual desktops is that users may connect to a different virtual machine each time they log in. Without a way to persist their settings and files, they'd start fresh every session. 

AVD solves this with a tool called FSLogix, which stores each user's profile settings, desktop preferences, and application data in a separate container on Azure file storage. When a user logs in, their profile loads automatically regardless of which virtual machine they're assigned to. This allows users to roam between session hosts without losing their settings. Profiles are stored on Azure file shares or Azure NetApp Files (higher performance). Configure storage access controls so only the relevant user and admins can access each profile container.

Azure prerequisites before deploying Azure Virtual Desktops

Before you provision your virtual desktop environment, a few things must already be in place and documented. Don’t skip these steps and try to configure them retroactively.

  • Azure Key Vault configured for secrets and encryption key management
  • Azure Firewall deployed and configured
  • Network Security Groups applied to all subnets
  • Microsoft Sentinel connected and receiving logs
  • Storage accounts secured with appropriate access controls
  • Subscription-level security configurations applied (Defender for Cloud, security policies)

These aren't just prerequisites for Azure Virtual Desktops, they're also CMMC controls that assessors will evaluate. Having them in place before deployment means your environment is compliant from day one rather than you trying to play catch-up later.

Mobile Device Management (MDM)

MDM is the endpoint strategy for companies with users who need to access CUI on physical devices. Either virtual desktops don't fit their workflow, they're field-based, or their role involves physical equipment that requires local software.

An MDM solution that's FedRAMP Moderate authorized enforces CMMC-required device configurations automatically when a device is enrolled. This includes:

  • Full disk encryption
  • OS patching and update management
  • Password/PIN policies
  • Screen lock and auto-wipe after failed authentication attempts
  • Application management and restriction
  • Certificate deployment for authentication
  • Remote wipe capability

The key word is "automatically." A good MDM solution doesn't require your IT team to manually interpret each NIST control and translate it into a device policy. Configurations are pushed at enrollment and maintained continuously, with drift detection and alerts when a device falls out of compliance.

MDM is generally a lower-cost alternative to virtual desktops, which makes it particularly relevant for smaller subcontractors, manufacturers, and field-based teams. The tradeoff is that enrolled physical devices are in CMMC scope, which means you'll need to document them in your SSP and maintain ongoing evidence of their compliance posture.

Choosing the right endpoint solution

Virtual Desktops (VDI) Federal MDM
CUI access method Cloud-hosted virtual desktop Physical device, MDM-hardened
Physical devices in scope? No Yes
Best for Office/remote knowledge workers Field teams, manufacturers, physical equipment users
Key requirement Reliable internet, <70ms latency FedRAMP Moderate authorized MDM
Mobile/iOS/Android support Supported with Intune MAM configuration (hybrid enforcement) Yes

When to choose virtual desktops

Virtual desktops are ideal for remote and hybrid teams, knowledge workers who handle CUI regularly, and organizations that want to keep their physical device fleet entirely out of CMMC scope. They're also a good choice for teams without dedicated IT staff to manage and maintain endpoint hardening across dozens or hundreds of devices.

The primary constraint is latency. VDI experience degrades significantly above roughly 70ms latency. Before deploying, test connection speeds from the worst-case user location: a remote employee on a slow connection is where VDI falls apart, and latency is usually the top user experience complaint. Choose Azure Government data center regions that minimize latency for your team's geography.

When to choose MDM 

MDM is the better fit when your users need to work on physical devices and virtual desktops aren't practical. This includes field engineers and manufacturing staff who operate equipment that requires local software, teams that rely on iOS or Android devices for CUI access (where VDI's screen capture protection requires additional Intune MAM configuration), and organizations where the per-seat cost of virtual desktops doesn't make sense for the number of CUI users involved.

MDM is also the right call if your users are frequently in locations with poor or inconsistent internet connectivity. Latency that would make a virtual desktop session unusable is a non-issue when CUI lives on a hardened physical device.

One requirement to plan around: your MDM solution must be FedRAMP Moderate authorized. Standard commercial MDM tools, even enterprise-grade ones, aren't built to meet federal compliance requirements out of the box. Verify FedRAMP authorization before you commit to a solution, and look for FedRAMP High if you also handle ITAR data.

When to choose a hybrid approach

In practice, many organizations use both. Office workers who handle CUI regularly get virtual desktops. Field engineers or manufacturing staff who can't practically use VDI get MDM-enrolled physical devices. The boundary is maintained in both cases: CUI stays within the enclave and is accessed only through controlled, documented endpoints.

If you're using both approaches, document both clearly in your SSP, including which users use which access method and how the boundary is enforced in each case.

Step 5: Identity architecture

We discussed above which identity controls you need to implement (MFA, RBAC, conditional access, etc.) But how do you actually structure Microsoft identity to support those controls? For organizations already running Microsoft 365 commercial, this is where things get complicated. A CMMC enclave on GCC High can’t just be added to your existing Microsoft environment; it has to be completely separate. 

If you're deploying Azure Virtual Desktop on GCC High, you have three options for how you manage identity:

  1. Users in Windows Server AD, virtual machines joined to Windows Server AD. Windows Server Active Directory (AD) is Microsoft's traditional on-premises identity system, and it's what most organizations have been using for years to manage user accounts, computers, and access policies. If you already have domain controllers (the servers that run AD) in Azure, this is usually the path of least resistance.
  2. Users in Windows Server AD, virtual machines joined to Azure AD Domain Services. Azure AD Domain Services is a managed domain service, meaning Microsoft runs the domain controllers for you. This eliminates the overhead of managing domain controller servers but requires synchronization between your on-premises or cloud Windows Server AD and Azure AD Domain Services.
  3. Users and virtual machines both in Azure AD Domain Services (fully cloud-native). The simplest option from a management perspective, and the right starting point for organizations building a new enclave from scratch without legacy AD infrastructure.

Tenant separation

In Microsoft's world, a "tenant" is your organization's dedicated instance of Microsoft 365, and it’s where your user accounts, email, files, and settings live. When you sign up for Microsoft 365, Microsoft creates a tenant for you. If you already have a Microsoft 365 commercial tenant account, your GCC High enclave needs to be a completely separate one. There is no upgrade path from commercial to GCC High, they're distinct environments on separate infrastructure.

This means your CUI users will have two Microsoft accounts: their commercial account (for everyday business tools) and their GCC High account (for CUI work). You’ll have to manage these carefully:

  • Commercial accounts should have no access to GCC High resources
  • GCC High accounts can be configured to access some commercial resources (read-only SharePoint, OneDrive, federated Teams chat), but only in one direction
  • Dual licensing is required for swivel-seat users. Verify licensing requirements carefully as these can change; Microsoft's licensing documentation is the authoritative source.

Step 6: Configuring your enclave to meet CMMC Level 2 requirements

CMMC Level 2 requires compliance with the 110 NIST 800-171 security controls. Not all of the control families are equally affected by your enclave architecture decisions, but several are almost entirely determined by how your enclave is designed and configured. 

Note that this is a working map to help you understand which controls your architecture decisions satisfy and doesn’t replace NIST 800-171 documentation or guidance from your C3PAO. 

Access Control (AC): 22 requirements

Access control is the largest family and the one most directly shaped by your enclave architecture. Key requirements include:

  • Limit system access to authorized users and the functions they're permitted to execute. This is your RBAC implementation: user roles defined in Entra ID, applied at the resource level, and reviewed on a regular schedule.
  • Control the flow of CUI in accordance with approved authorizations. Network segmentation and DLP work together to satisfy this requirement. On the network side, NSG rules and Azure Firewall policies define which systems can send data where. At the application layer, Purview DLP policies enforce that CUI-labeled content can only flow to authorized destinations. For example, blocking a GCC High SharePoint file from being shared externally via a public link, even if the user attempts it manually.
  • Separate duties of individuals to reduce the risk of malevolent activity. Role separation in practice means that the person who administers your Azure environment shouldn't also be the one approving their own access changes. Define at least two distinct privileged roles (system administration and security oversight) and enforce that separation in Entra ID.
  • Employ the principle of least privilege, including for privileged accounts. Admin accounts should only exist in the enclave environment, should only be used for admin tasks, and should be separate from the user's day-to-day account. Even your IT administrator should have a standard account for everyday work and a separate privileged account for enclave management.
  • Use non-privileged accounts for non-privileged activities
  • Prevent non-privileged users from executing privileged functions (admin access controls)
  • Control remote access sessions. Every remote session into the enclave must use MFA, must travel over an encrypted connection, and must terminate after a defined period of inactivity. Configure session timeout policies in AVD and conditional access policies in Entra ID.
  • Control wireless access to organizational systems (relevant for hybrid enclaves)
  • Control connection of mobile devices. Every device that accesses CUI must be enrolled in MDM before it can connect. Conditional access policies in Entra ID can enforce this automatically, so if a device isn't enrolled and compliant, it can't authenticate to enclave resources.
  • Encrypt CUI on mobile devices and mobile computing platforms
  • Verify and control any connections to external systems

Your enclave architecture directly addresses many of these through network segmentation, RBAC configuration, conditional access policies, and endpoint controls. Document each implemented control in your SSP with a description of how your specific configuration satisfies the requirement.

Audit and Accountability (AU): 9 requirements

  • Create and retain system audit logs to enable monitoring, analysis, investigation, and reporting of unlawful or unauthorized activity
  • Ensure that the actions of individual users can be traced to those users
  • Review and update logged events
  • Alert in the event of an audit process failure
  • Correlate audit record review, analysis, and reporting processes
  • Provide audit record reduction and report generation
  • Protect audit information and tools from unauthorized access, modification, and deletion
  • Limit management of audit logging to a subset of privileged users
  • Collect audit information into one or more central repositories

Your Sentinel configuration is the primary mechanism for this. Microsoft-hosted service logs from Azure AD, Defender, and Office 365 feed into Sentinel automatically and at no ingestion cost, but you need to actively configure what gets collected, how long it's retained, and who can access it. 

Define your log retention period in a data retention policy (the requirement is that logs support after-the-fact investigation; 90 days accessible and one year archived is a common baseline), enforce that retention technically in Sentinel, and restrict log management to a named set of privileged administrators. 

Configuration Management (CM): 9 requirements

  • Establish and maintain baseline configurations and inventories of organizational systems
  • Establish and enforce security configuration settings
  • Track, review, approve, and log changes to organizational systems
  • Analyze the security impact of changes prior to implementation
  • Define, document, approve, and enforce physical and logical access restrictions associated with changes
  • Employ the principle of least functionality (disable unnecessary ports, protocols, functions)
  • Restrict, disable, or prevent the use of nonessential programs, functions, ports, protocols, and services
  • Apply deny-by-exception (blacklisting) or allow-by-exception (whitelisting) policies for software
  • Control and monitor user-installed software

This family is largely about documentation and process, not just configuration. Your AVD base image represents a baseline configuration, meaning it should be documented, version-controlled, and changed only through a formal process. Your NSG rules and firewall policies are also baseline configurations that need to be inventoried and reviewed. The key test: if an assessor asks to see your baseline configuration for any enclave system, can you produce a current, accurate document? If not, that's a gap.

Identification and Authentication (IA): 11 requirements

  • Identify system users, processes acting on behalf of users, and devices
  • Authenticate the identities of users, processes, or devices as a prerequisite to accessing CUI systems
  • Use multifactor authentication for local and network access to privileged accounts and for network access to non-privileged accounts
  • Employ replay-resistant authentication mechanisms
  • Prevent reuse of identifiers for a defined period
  • Disable inactive accounts
  • Enforce a minimum password complexity
  • Prohibit password reuse for a specified number of generations
  • Allow temporary password use with immediate change requirement
  • Store and transmit only cryptographically-protected passwords

Your Entra ID configuration is the foundation for this entire family. Configure MFA enforcement through conditional access policies so it applies to all users accessing the enclave, not just admins. Set password complexity and history requirements in Entra ID's authentication policies. Configure an account inactivity policy (35 days is a common threshold) and automate account disabling for accounts that haven't logged in. The critical point is that MFA must be enforced at the platform level, not left to individual users to configure.

System and Communications Protection (SC): 16 requirements

This family governs network architecture and encryption, and it's where your firewall configuration, NSGs, and encryption key management are evaluated.

  • Monitoring, controlling, and protecting communications at external boundaries and key internal boundaries. In Azure Government, this means your Azure Firewall is logging and inspecting all traffic entering and leaving the enclave, and your NSGs are controlling traffic between internal subnets. 
  • Implementing architectural designs, software development techniques, and systems engineering principles that promote effective information security. Your network topology should be documented and reviewed as part of your SSP.
  • Separating user functionality from system management functionality
  • Preventing unauthorized and unintended information transfer
  • Implementing cryptographic mechanisms to prevent unauthorized disclosure of CUI during transmission
  • Terminating network connections after defined periods of inactivity

Your Azure Firewall configuration, NSG policies, VPN encryption standards, and Key Vault setup are the primary technical mechanisms here. FIPS-validated cryptography is required for protecting CUI, so verify that all encryption implementations (at rest and in transit) use FIPS-validated modules.

System and Information Integrity (SI): 7 requirements

  • Identify, report, and correct system flaws in a timely manner (patching)
  • Provide protection from malicious code at appropriate locations
  • Monitor organizational systems to detect attacks and indicators of potential attacks
  • Identify unauthorized use of organizational systems
  • Receive and respond to threat intelligence from information-sharing forums and sources

Microsoft Defender and your Sentinel alert rules are the primary mechanisms here. Defender provides malware protection and threat detection across your virtual machines, endpoints, and cloud services. Sentinel aggregates those signals and lets you configure alert rules that fire when suspicious activity is detected. For example, a user downloading an unusually large number of files, or a login from an unexpected geographic location. 

For cloud workloads, Microsoft handles OS-level patching at the infrastructure level, but you're responsible for patching any applications running on your virtual machines.

Step 7: Preparing documentation for a C3PAO assessment

A well-architected enclave that isn't properly documented will still fail a C3PAO assessment. Your documentation needs to accurately describe your environment as it actually operates, not as you intended to configure it, and not as it was configured six months ago.

System Security Plan (SSP)

The SSP is the central document of your CMMC compliance program. For your enclave, it must clearly document:

  • The system boundary: which systems, users, and applications are inside the enclave
  • Which systems are explicitly out of scope and why
  • How the boundary is technically enforced (firewalls, NSGs, access controls, monitoring)
  • How CUI flows through the environment
  • For each of the 110 NIST 800-171 controls: whether it's implemented, partially implemented, or not implemented, with a description of how it's satisfied (or planned)

A vague SSP is one of the most common assessment preparation failures. "We use Azure" is not an implementation statement. "Entra ID conditional access policies enforce MFA for all users accessing CUI systems via [specific policy name], applied to the [tenant name] GCC High environment" is an implementation statement.

Your SSP should be written so that an assessor can evaluate your controls without having to ask you questions. If they have to ask, it's either not documented or the documentation doesn't match the environment.

Recommended reading

How to Write a System Security Plan for CMMC + SSP Template

Plan of Action and Milestones (POA&M)

The POA&M documents any controls that aren't yet fully implemented, along with your plan and timeline for addressing them. Having gaps isn't necessarily disqualifying, but those gaps need to be documented honestly and accompanied by specific and realistic remediation timelines.

Your POA&M is directly tied to your SPRS score. The Supplier Performance Risk System (SPRS) is the DoD's scoring system for defense contractor cybersecurity. Your score starts at 110 (one point for each of the 110 NIST SP 800-171 requirements) and is reduced based on gaps. Different requirements carry different point weights depending on their importance, so a single unimplemented requirement can reduce your score by anywhere from one to five points. 

Your SPRS score is self-reported and must be submitted to the DoD's SPRS database. CMMC Level 2 requires a provisional SPRS score of at least 88 before you can proceed to a C3PAO assessment. Understand your score and any gaps before you engage a C3PAO.

Recommended reading

SPRS Scoring: How to Get a Current CMMC Status and Stay Eligible for DoD Contracts

Required security policies and procedures

Assessors will ask for written policies covering how your organization manages your enclave. At minimum, you'll need documented policies covering:

  • Acceptable Use Policy (AUP) for enclave systems
  • Access Control Policy (how access is granted, reviewed, and revoked)
  • Incident Response Policy and Plan (specific to your enclave)
  • Configuration Management Policy (how changes are managed and documented)
  • Audit Log Review procedures
  • Media Protection Policy (handling of removable media and printed CUI)
  • Personnel Security procedures (user training, background check requirements, account termination)

Assessors will not accept boilerplate policies or generic templates. Make sure your documents describe your specific environment and procedures that reflect your real practices. 

Enclave-specific assessment evidence 

Your SSP and policies describe what you do, and evidence proves you do it. Before your assessment, you should be able to produce:

  • Configuration exports from your cloud environment (Entra ID policies, firewall rules, MDM configurations)
  • Audit logs demonstrating that logging is active and collecting the required events
  • Evidence of MFA enrollment for all enclave users
  • Patch compliance reports for all in-scope systems
  • User access reviews showing current RBAC assignments
  • Completed user training records
  • Incident response test documentation

Collect this evidence continuously, not as a last-minute scramble before the assessment. Time-stamped, automatically collected evidence is significantly more credible to assessors than manually gathered screenshots that are likely out of date.

Build or buy? Choosing a DIY enclave vs. a managed solution

Now that you have a clearer picture of what goes into a CMMC-compliant enclave, the next question is: should you build and maintain your own enclave, or opt for a managed solution? Let’s take a closer look at each option.  

Building a secure enclave

Building a CMMC enclave from scratch is a substantial IT project. A DIY build requires:

  • Azure Government expertise: Setting up a GCC High tenant, configuring Azure infrastructure, deploying virtual desktops, and connecting everything properly requires fluency with Microsoft's government cloud environment, which is meaningfully different from commercial Azure.
  • Security engineering: Configuring Sentinel, deploying Defender, setting up conditional access policies, writing NSG rules, and configuring Key Vault correctly takes time and expertise.
  • CMMC-specific knowledge: Knowing which Azure configurations satisfy which NIST controls, and how to document them for assessor review, requires familiarity with both technical and compliance requirements.
  • Ongoing maintenance: The enclave doesn't maintain itself. Patches, configuration updates, new user provisioning, evidence collection, and drift detection are ongoing responsibilities.
  • Documentation: Your SSP, POA&M, and policies need to accurately reflect your live environment at all times. Manual documentation quickly drifts from reality.

For organizations with a dedicated IT security team and strong Microsoft cloud expertise, a DIY build is feasible. For most small and mid-size contractors where IT is one person or a part-time function it's a significant undertaking. A misconfigured enclave creates a false sense of compliance and can fail assessment in ways that are difficult to diagnose and remediate.

Timelines are also a real factor. Building and validating a CMMC enclave from scratch typically takes months of focused effort. If your contracts are tied to specific compliance deadlines, that timeline matters.

Opting for a managed solution

A managed enclave solution like Secureframe Defense handles the infrastructure build and configuration for you, provides compliance automation tooling on top of it, and maintains the environment over time.

  • Infrastructure: Your environment is deployed from a pre-configured baseline with the required CMMC security controls enforced from the start, so you're not starting from zero on control implementation.
  • Documentation: Your SSP is automatically generated from your live environment, not written from a blank template. Controls are documented based on actual configuration.
  • Compliance management: Instead of tracking 110 controls in a spreadsheet and figuring out which requirements apply to you or what to do first, a guided workflow breaks your compliance journey into clear, prioritized steps.
  • Drift detection: Continuous monitoring flags deviations from your compliant baseline as they happen, so you can identify and fix any gaps before an assessor does. 
  • Assessment readiness: Evidence is automatically collected from your environment and validated so you're ready whenever your C3PAO window opens, not scrambling in the weeks before.

For most small and mid-size contractors — especially those trying to get certified quickly to maintain contract eligibility — the build vs. buy decision usually comes down to time and expertise. For teams without deep cloud security knowledge, a managed approach gets you assessment-ready faster and reduces the risk of costly misconfigurations.

Keeping your enclave compliant over time

Building your enclave isn’t the finish line. CMMC requires continuous compliance, and the compliance posture you demonstrated during your assessment will start to erode the moment you stop actively maintaining it.

Teams get new employees and forget to provision enclave accounts correctly. Software gets updated and firewall rules change. A well-intentioned IT change loosens a security configuration. A new application gets added to the environment without going through the proper change management process. If that drift goes undetected until your next assessment cycle, you’ll have a problem.

Continuous monitoring and drift detection

Configuration drift is what happens when your live environment diverges from your documented compliant baseline. It's almost always unintentional and almost always happens gradually.

Manual periodic audits are not sufficient. By the time you do a quarterly review, the drift has already happened. What you need is automated monitoring of your live environment against your documented baseline, with alerts when something changes.

Change management

Every change to your enclave needs to go through a documented change management process, no matter how small. You’ll need: 

  • A defined process for requesting, reviewing, approving, and documenting changes
  • Security impact analysis before significant changes are implemented
  • Documentation updates that reflect the changed configuration
  • Post-change verification that the environment still meets baseline

This process doesn't need to be bureaucratic. For a 50-person contractor, your change management procedure can be a simple Jira ticket or a form-based request. What matters is that it's documented, followed consistently, and that your SSP reflects your current environment.

Adding users and changing boundaries

Onboarding a new user who needs CUI access is a compliance event, not just an IT ticket. It requires:

  • Provisioning a new enclave account (separate from commercial)
  • Assigning appropriate RBAC roles
  • Enrolling the user's endpoint in MDM or provisioning a virtual desktop
  • Completing user training and documenting completion
  • Updating your SSP if the new user's role or access pattern changes your system boundary documentation

Similarly, any change to your enclave boundary (adding a new system, integrating a new application, or connecting a new physical location) requires a formal boundary review and SSP update before the change is implemented.

Collecting evidence between assessments

Your CMMC certification is valid for three years, but you should be collecting evidence continuously, not rebuilding your evidence package from scratch when your next assessment approaches. Continuous evidence collection:

  • Reduces assessment preparation time dramatically
  • Gives you real-time visibility into your compliance posture
  • Provides a documented history that supports your SSP claims
  • Makes it easy to respond quickly when your C3PAO requests specific evidence

How Secureframe Defense simplifies the process

The full scope of building and running a defensible CMMC enclave involves complex architecture decisions, configuration requirements, documentation, and ongoing compliance maintenance. It’s a significant undertaking, requiring dedicated time, resources, and technical expertise. 

Secureframe Defense is built to handle CMMC end to end, including a compliant enclave, so you don't have to piece it all together yourself.

Automated cloud provisioning

Secureframe Defense automatically provisions a CMMC-aligned cloud enclave in your choice of Microsoft 365 GCC High or Google Workspace for Government. Logging, monitoring, access control, and network segmentation are enforced from day one. There's no manual Azure configuration, no working through hundreds of pages of Microsoft documentation, and no risk of deploying an environment and then hardening it later.

The enclave is your environment and you own the tenant, but the configuration comes pre-built against CMMC requirements rather than starting from scratch.

Virtual Desktops and Federal MDM

Once your enclave is provisioned, Secureframe Defense enables users to securely access it.

Virtual Desktops are provisioned directly from the Secureframe platform. Each virtual desktop runs in Azure Government and is pre-configured to meet CMMC Level 2 technical requirements. No Azure Government expertise required, no manual configuration of AVD enforcement settings. Your team gets a familiar Windows 11 desktop experience, and their physical devices stay out of CMMC scope.

Secureframe Federal MDM is a FedRAMP Moderate authorized MDM solution for organizations that need to secure physical endpoints. When a device is enrolled, Secureframe automatically pushes all required CMMC Level 2 configurations, including encryption, patching, access controls, and configuration management, with no manual engineering effort. Compliance is continuously enforced, with drift detection and clear remediation guidance when a device falls out of configuration.

Both options tightly control CUI access, log all activity, and feed compliance evidence into your Secureframe Defense environment automatically.

Defense Navigator

Defense Navigator is the guided workflow that takes you from initial scoping to assessment readiness. It walks you through structured scoping questions, applies the relevant CMMC requirements to your specific environment, and breaks the 110 controls into clear, prioritized tasks. Your SPRS score updates in real time as you complete steps, so you always know where you stand.

Guidance is built by former federal assessors with direct CMMC Level 2 certification experience. You're following a proven path, not figuring it out as you go.

Automated Documentation

Once your enclave and integrations are in place, Secureframe generates your SSP, POA&M, and required policies directly from your live environment. Control descriptions and implementation statements are built from actual configuration data, not copied from boilerplate templates that may or may not reflect your real environment.

As your environment evolves, your documentation automatically stays current. Configuration changes are reflected in your SSP, and new and remediated gaps are updated in your POA&M. The documentation your assessor reviews matches the environment they're evaluating.

Continuous Compliance Monitoring

After certification, Secureframe Defense continuously monitors your enclave for configuration drift and compliance gaps. Automated alerts flag deviations before they become assessment findings. Evidence of device compliance, control implementation, and configuration health flows continuously into your compliance record, so your next assessment isn't a rebuild, it's a checkpoint on a program you've been actively managing.

Assessment-ready documentation

When you're ready to bring in a C3PAO, Secureframe exports a complete, properly formatted assessment package: all test evidence, SSP documentation, and the eMASS pre-assessment form. Assessors get everything they need in the format they expect, without back-and-forth or last-minute documentation fixes.

Secureframe is also a Registered Practitioner Organization and among the first organizations to achieve CMMC Level 2 certification, which means our platform is built by people who have been through the process firsthand.

Ready to see it in action? Schedule a demo or visit secureframe.com/cmmc to learn more.

Provision your CUI enclave

Request a demo

FAQs

What is a CMMC enclave? 

A CUI enclave is a secure, isolated environment designed to store, process, and transmit CUI data. It helps organizations in the Defense Industrial Base (DIB) meet requirements under frameworks like NIST 800-171 and the Cybersecurity Maturity Model Certification (CMMC) by limiting the scope of systems that must be secured—making compliance more scalable and easier to manage.

How does a secure enclave help with CMMC compliance? 

CMMC Level 2 assessments evaluate everything in scope. The more systems that touch CUI, the larger your compliance footprint: more assets to configure, more evidence to produce, more remediation work. An enclave reduces your scope to a manageable set of systems, which lowers cost, complexity, and assessment risk.

What's the difference between a CUI enclave and going enterprise-wide? 

An enterprise-wide approach applies all 110 NIST SP 800-171 controls to your entire IT environment. An enclave applies those controls only to the isolated boundary where CUI lives, keeping the rest of your environment out of scope. Enclaves are more cost-effective when a relatively small percentage of your workforce handles CUI. Enterprise-wide compliance makes more sense when CUI is pervasive across your organization.

How long does it take to build a CMMC enclave? 

A DIY build typically takes several months for a team without prior experience. With an automated provisioning solution, the infrastructure can be stood up in hours, with the remaining timeline driven by documentation, control implementation, and assessment scheduling. Industry average preparation time is six to eighteen months; organizations using automated solutions can typically reach readiness significantly faster.

Can I use physical devices with a CUI enclave? 

Yes, through two approaches. Virtual desktops keep physical devices entirely out of scope by ensuring CUI never leaves the cloud. MDM-enrolled physical devices can also be used to access CUI while maintaining a controlled boundary. The devices are in scope, but MDM ensures they're continuously configured to CMMC requirements. Some organizations use both, depending on user role and workflow.

What is the difference between a managed enclave and a hosted enclave?

In a managed enclave model, you own the Microsoft tenant and Azure subscription. The environment persists if you change vendors, scales with Microsoft's infrastructure, and you're not paying per-seat to a third party for hosting. In a hosted enclave model, a vendor owns the hardware and you pay per-seat billing. Managed enclaves offer lower vendor lock-in risk and greater data portability; hosted enclaves may require months to migrate away from if you change providers.

Emily Bonnie

Senior Content Marketing Manager

Emily Bonnie is a seasoned digital marketing strategist with over ten years of experience creating content that attracts, engages, and converts for leading SaaS companies. At Secureframe, she helps demystify complex governance, risk, and compliance (GRC) topics, turning technical frameworks and regulations into accessible, actionable guidance. Her work aims to empower organizations of all sizes to strengthen their security posture, streamline compliance, and build lasting trust with customers.