Microsoft infrastructure architecture considerations: part 3 (controlling network access)

Continuing the series of posts on the architectural considerations for designing a predominantly-Microsoft IT infrastructure, based on the MCS Talks: Enterprise Infrastructure series, in this post, I’ll look at some of the considerations for controlling access to the network.

Although network access control (NAC) has been around for a few years now, Microsoft’s network access protection (NAP) is new in Windows Server 2008 (previous quarantine controls were limited to VPN connections).

It’s important to understand that NAC/NAP are not security solutions but are concerned with network health – assessing an endpoint and comparing its state with a defined policy, then removing access for non-compliant devices until they have been remediated (i.e. until the policy has been enforced).

The real question as to whether to implement NAC/NAP is whether or not non-compliance represents a business problem.

Assuming that NAP is to be implemented, then there may be different policies required for different groups of users – for example internal staff, contractors and visitors – and each of these might require a different level of enforcement; however, if the the policy is to be applied, enforcement options are:

  • DHCP – easy to implement but also easy to avoid by using a static IP address. It’s also necessary to consider the healthcheck frequency as it relates to the DHCP lease renewal time.
  • VPN – more secure but relies on the Windows Server 2008 RRAS VPN so may require a third party VPN solution to be replaced. In any case, full-VPN access is counter to industry trends as alternative solutions are increasing used.
  • 802.1x – requires a complex design to support all types of network user and not all switches support dynamic VLANs.
  • IPSec – the recommended solution – built into Windows, works with any switch, router or access point, provides strong authentication and (optionally) encryption. In addition, unhealthy clients are truly isolated (i.e. not just placed in a VLAN with other clients to potentially affect or be affected by other machines). The downside is that NAP enforcement with IPSec requires computers to be domain joined (so will not help with visitors or contractors PCs) and is fairly complex from an operational perspective, requiring implementation of the health registration authority (HRA) role and a PKI solution.

In the next post in these series, I’ll take a look at some of the architectural considerations for using virtualisation technologies within the infrastructure.

Leave a Reply