Hi 宋宇杰,
How is your issue going? Has it been resolved yet? If it has, please consider accepting the answer as it helps others sharing the same problem benefit too. Thank you :)
VP
This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
The specific situation is as follows:
The main questions are:
How should I choose between Windows Server 2019, 2022, and 2025?
Hi 宋宇杰,
Your assumption that upgrading forest and domain functional levels directly breaks Windows XP and Windows 7 compatibility is a common misconception. Functional levels govern backend directory features and strictly dictate which server operating systems are permitted to act as Domain Controllers, not which client systems can authenticate. The actual threat to legacy client connectivity comes from default security protocols on newer server operating systems, such as the deprecation of SMBv1 and RC4 Kerberos encryption. For Windows XP to authenticate against newer domain controllers, you must often ensure RC4 is enabled via the Network security: Configure encryption types allowed for Kerberos policy, located at Computer Configuration\Windows Settings\Security Settings\Local Policies\Security Options. You can safely raise your functional level to Windows Server 2016 since your existing servers run that operating system, though maintaining these outdated protocols to support legacy clients introduces significant security vulnerabilities to your infrastructure.
When selecting the operating system for the new Huawei Cloud server, Windows Server 2025 is completely unsupported unless you raise the functional levels, as it enforces a strict minimum functional level requirement of Windows Server 2016. If you remain on the 2008 R2 functional level, your valid options outside of 2016 are Windows Server 2019 and Windows Server 2022. Windows Server 2022 is the optimal choice for your environment, as it offers the longest remaining lifecycle and seamlessly integrates into a 2008 R2 functional level while maintaining compatibility with legacy clients, provided the necessary legacy encryption types and protocols are manually permitted.
Allocating hardware resources for a domain controller requires focusing on memory and storage integrity rather than raw processor speed. Active Directory attempts to load the entire database file, located at C:\Windows\NTDS\ntds.dit, into system RAM to process queries instantly without relying on slower disk reads. You should provision the virtual machine with at least two virtual CPUs and enough memory to exceed the size of your current database, with four to eight gigabytes serving as a typical starting point for medium environments. It is highly recommended to provision a separate virtual disk dedicated exclusively to the NTDS database, transaction logs, and the SYSVOL folder. Disabling host-level write caching on this secondary data disk is absolutely crucial to prevent Active Directory database corruption during an unexpected hypervisor reboot or cloud failure.
Preparing your cloud environment before deployment requires establishing secure routing and defining the logical network topology. You must first configure a persistent site-to-site VPN or direct interconnect between Huawei Cloud and your existing environments to guarantee continuous replication traffic. Next, assign a static IP address to the new server and configure its primary DNS server to point to one of your existing local or Azure Domain Controllers, allowing it to locate the domain during the promotion process. Finally, use the Active Directory Sites and Services console to create a new site representing the Huawei Cloud location, bind the new cloud network subnets to it, and configure site links to ensure future cloud workloads authenticate locally rather than traversing the wide area network back to your primary data center.
Hope this answer brought you some useful information. If it did, please hit “accept answer”. Should you have any questions, feel free to leave a comment.
VP
From the provided information:
Key points from the context:
Relevant excerpts:
Given that the current forest/domain functional level is Windows Server 2008 R2, the context only explicitly describes support matrices for 2008, 2008 R2, 2012, 2012 R2, 2016, and 2025, and explicitly states that Windows Server 2025 DCs require at least Windows Server 2016 functional level. It does not provide a complete compatibility matrix for adding Windows Server 2019 or 2022 DCs into a 2008 R2 forest.
From the context, the only fully documented and safe pattern is:
Because the current forest/domain functional level is Windows Server 2008 R2 and must be retained, and the context does not explicitly state that Windows Server 2019 or 2022 domain controllers can be added to a 2008 R2 forest, the only fully supported guidance in the context is to continue using a DC OS that is already known to work in the current environment: Windows Server 2016.
Therefore, based strictly on the provided documentation, the recommended and documented-safe choice is to deploy the new domain controller on Windows Server 2016, matching the existing DCs and the current functional level behavior described.
The context does not provide any concrete sizing guidance (CPU, RAM, disk, or network throughput) for domain controllers. It only describes functional levels, deployment steps, and upgrade patterns. No numeric or qualitative sizing rules are given.
Therefore, no specific recommendations for CPU, memory, disk, or network allocation can be derived from the provided material.
From the context on functional levels:
And from the article on raising functional levels:
To avoid conflicts or incompatibility when choosing an OS for a new domain controller, the context supports these principles:
Applied to this scenario, and constrained by the context:
The context describes general patterns for adding a new domain controller and upgrading domain controllers:
And for new DCs in Azure (which is conceptually similar to adding a DC in another cloud):
From these, the preparatory steps supported by the context are:
Because the context does not provide more detailed pre-deployment checks (such as site topology, replication, or capacity planning), only these high-level preparations can be stated.
References: