In the recent blogs about DHCP server in Windows Server 2012, we discussed DHCP Failover – a new mechanism for achieving high availability of the Windows DHCP server and DHCP PowerShell – the new command line interface for managing Windows DHCP server. This blog discusses another new feature in DHCP Server in Windows Server 2012 – DHCP Policies. This feature is also referred as Policy based IP address and option assignment or just Policy Based Assignment (PBA) for short.
Envision the following scenarios that are faced by administrators:
- Your subnet has a mix of different types of clients – desktop computers, printers, IP phones etc. You would like different types of clients to get IP addresses from different IP address ranges within the subnet. For example, IP phones should get IP addresses from the range 10.10.10.10 - 10.10.10.50 (in the 10.10.10.0/24 subnet), a different TFTP server and bootfile name option.
- In a subnet which has a mix of wired and mobile computers, you may want to assign shorter lease durations to mobile computers (e.g. 4 hours) and longer lease durations to wired computers (e.g. 4 days)
- Your deployment requires that you control who gets access to your network i.e. you want to provide DHCP service to a set of known clients (based on MAC address) for each subnet.
- With employees bringing in their own devices (smartphones, tablets) at work, you want to handle network traffic or control network access based on type of device.
DHCP Policies provide the DHCP admin with a very useful lever to achieve these scenarios.
As of Windows Server 2008 R2, an admin configures an IP address range and option values for a scope/subnet. All clients which are in that scope/subnet get an IP address from this IP address range of the scope and get options configured for the scope. If an administrator of the DHCP server needs to further apportion the IP address range of a scope to be delivered to a specific class of clients or devices or needs to give out different option values to different types of clients - as the aforementioned scenarios demand – there was no way for an admin to achieve that (unless you used individual reservations, which are effort intensive to manage). So, essentially, granularity at which you could assign IP addresses and options existed only up to the scope level. The DHCP policies in Windows Server 2012 help the administrator achieve exactly that – a more granular mechanism to assign IP addresses and options.
With Windows Server 2012, you can create policies on the DHCP server. A policy consists of 2 main parts – conditions and settings. Condition(s) specified in a policy allows you to group clients. Settings are the network configuration parameters (IP address, options, lease duration) that are provided to the clients in the DHCP server response. Conditions can be specified based on a set of fields which are present in the DHCP client request. The policy model is fairly simple: every DHCP client request is evaluated against the conditions in a policy. If a client request matches the conditions in the policy, the settings associated with a policy will be provisioned to the client via the DHCP server response.
In Windows Server 2012, you can specify five different criteria (a fixed set) based on which one can segregate or group clients:
The operators that can be used in the conditions are equals and not equals. You can also use a trailing wild card with MAC address, Vendor Class, User Class and Client identifier conditions to perform a partial match. Combine the equals or not equals with a wild card in the value specified in the condition and you effectively achieve a starts with or does not start with condition.
You can either have a single condition in a policy or a set of conditions which can be ORed or ANDed. For example, "Vendor Class Equals Cisco IP Phone 7940" is a condition (Cisco IP Phone 7940 is the value of Vendor Class for Cisco IP Phone version 7940). Also a grouping such as "User Class Equals LabComputers" AND "MAC Address Not Equals 00-11-22*" is a group of two conditions. Each policy is created with either a single condition or a set of such conditions. An incoming client requesting for an IP address and options from the DHCP server is said to satisfy a policy if the client satisfies the cumulative set of conditions in the policy.
A client which does not match conditions of any policy, is leased an IP address from the rest of the IP address range of the scope (exclusive of all the policy IP address ranges) and given option values configured at the scope.
When a client matches the conditions of a policy, the DHCP server responds to the clients based on the settings of a policy. Settings associated to a policy can be an IP address range and/or options. An administrator could configure the policy to provide an IP address from a specified sub-range within the overall IP address range of the scope. You can also provide different option values for clients satisfying this policy.
Policies can be defined server wide or for a specific scope. A server wide policy – on the same lines as server wide option values – is applicable to all scopes on the DHCP server. A server wide policy however cannot have an IP address range associated with it.
In addition to the IP address range and options which can be associated with a policy, there are two other noteworthy “settings” for a policy. You can set the IP address lease period for a policy. This allows you to configure a longer or shorter lease duration for clients which match the policy conditions. You can also configure how DNS registrations should be handled for clients matching the policy conditions. Any DNS registration behavior of the DHCP server which can be configured server wide or on a per scope basis – for example, turn on/off the DNS registration (and deregistration) or DNS name protection – can be configured on a per policy basis.
A policy can have only an IP address range (no options) or only options (no IP address range) as a setting.
You can configure more than one policy within a scope or even server wide. Every policy has an admin assigned processing order. While processing client requests, the DHCP server evaluates the client requests against the conditions in the different policies based on the processing order of the policy – with processing order 1 policy being processed first. Scope level policies are processed first by the DHCP server followed by server wide policies.
If a client satisfies the conditions of more than 1 policy, it will get the settings in an aggregated manner from the different policies it satisfied. What this implies is if, for example - policy-1 has an option value for option 3 (router) and policy-2 has an option value for option 6 (DNS server) and if a client request matched the condition set of both policies, the server will respond with router value of policy-1 and DNS server value of policy-2. However, in this example, if policy-1 also had an option value for DNS server, the client will get both (router and DNS server) option values from policy-1 if policy-1 is higher up in processing order i.e. the DNS server option value in policy-2 is overridden.
A policy does not have to be configured with all option values that you have already configured at the scope level options. If a policy client has requested an option which is not present in the policy but has been configured in the scope level options, the scope level option value would be returned to the client in the server response.
There a couple of ways to segregate clients based on the type of device. One way to do this is by using vendor class/identifier. This string sent in option 60 by most DHCP clients identify the vendor and thereby the type of the device. Another way to segregate clients based on device type is by using the MAC address prefix. The first three bytes of a MAC address is called OUI and identify the vendor or manufacturer of the device.
By creating DHCP policies with conditions based on Vendor Class or MAC address prefix, you can now segregate the clients in your subnet in such a way, that devices of a specific type get an IP address only from a specified IP address range within the scope. You can also give different set of options to these clients.
After assigning IP addresses from a specific IP address range to a class of devices, you can configure your router to handle network traffic from this IP address range differently. In effect, you can achieve network access control for a class of devices. By configuring route options (default gateway - option id 3, classless static routes - option id 121) on a DHCP policy, you can also control flow of network traffic from specific types of devices.
Let’s say you need to configure shorter lease durations to Wi-Fi devices and longer ones to wired devices. Access Points (APs) are typically capable of behaving as a DHCP relay agent (or are connected to one) which can be configured with Option 82 - DHCP relay agent option. Presence of a specific value in the relay agent option in this case would indicate a Wi-Fi device. You can create a policy with a condition based on the relay agent information option value which segregates these WiFi clients and configure this policy for a shorter lease duration. The rest of the clients in the scope will continue to get the longer lease duration configured at the scope.
Another scenario, if you need to perform MAC-based filtering at the scope level, you could create a scope-level policy using MAC Address as the criteria in the condition.
There are several such scenarios which DHCP policies can cater to in terms of provisioning of IP addresses and settings. We will be talking more about some of these scenarios in upcoming blog posts.
You can find the step-by-step guide for configuring DHCP policies here.
In conclusion, DHCP policies in Windows Server 2012 enables grouping of clients/devices using the different criteria and delivering targeted network configuration to them. We hope that you will find policy based assignment useful to your deployment needs! We would love to hear any feedback you have to share on the same.