Microsoft Lync Server 2010 : Planning for Voice Deployment - Network Configuration, Call Admission Control

1. Network Configuration

Planning the network configuration first is key when beginning to plan for Call Admission Control, Media Bypass, or Enhanced 911 services in Lync Server 2010. Each of these components relies on network regions, sites, and links to be configured correctly before they can be enabled. The definition of each region, sites, and subnets might seem overwhelming at first, but it offers quite a bit of flexibility and control over call routing.

There are a few basic components that must first be understood when planning for these services:

• Regions— Network regions are the backbone of a network. Each network region must be associated with a Lync Server central site defined within the topology. This is a site where Lync Front End Servers are deployed and users are homed. Network regions are typically a hub where many other network sites are connected. Examples of regions are North America and Europe, or even on a smaller scale such as West Coast and East Coast.

• Sites— Each network region consists of at least one site and possibly many more. Sites are offices or locations that are associated with the major network region. In other words, all the offices or locations that have users homed in the central site for the region should be created as sites. This site definition should not be confused with the central site to which the network region was associated. In fact, a network site object should also be created for the central site if users or voice gateways are physically located there. This might seem redundant, but is required for appropriate routing.

• Subnets— Each site defined should contain at least one subnet definition. Each subnet used at a site should be entered and associated with the correct site. Lync endpoints are associated with a site and region by matching to a subnet defined here. The Call Admission Control and Media Bypass features rely on matching the subnet of the IP/PSTN gateway to callers, so be sure to include the subnets used for voice hardware.

• Bandwidth Policy Profiles— Bandwidth policy profiles define a network link speed and the available bandwidth for audio or video calls. Both the individual session limit (two-way) can be specified as well as the total amount of bandwidth used for audio and video traffic. Bandwidth policy profiles are associated with a site or region link. Sites do not require a bandwidth policy profile to be assigned. In fact, if sites within a region are not bandwidth constrained, no profile should be assigned. Assign bandwidth policy profiles only to sites that should have their audio and video WAN usage limited.

• Region Links— When multiple regions exist, and by definition a region is multiple Front End pools with different central sites, region links should be defined that identify the amount of bandwidth available between regions. When defining a region link two regions are required as well as a bandwidth policy profile to apply between the two regions.

• Region Route— A region route specifies how two regions should be connected. In many cases, a region route mimics a region link and can just be between two different regions. In other cases where two regions are not directly linked, but share a link to a common region, a region route defines how these regions must traverse the common region to communicate. In that case, two different region links must be crossed, which might each have a different bandwidth policy profile.

• Site Link— The final component of the network configuration is a site link. In most cases, sites are connected to a network region directly, which acts as a hub for the users. There might be instances where in addition to a connection to the network region central site, sites have a direct connection to each other that bypasses the central site. Site links are used to create these objects that can then have a bandwidth policy profile associated.

Figure 1 depicts a sample configuration. The definition of each component will vary based on the organization.

Figure 1. Network Configuration Example

2. Call Admission Control

Planning to deploy call admission control features in Lync Server 2010 is going to depend greatly on the network configuration discussed in the previous section. The basis for this is because this feature relies heavily on locating an endpoint based on subnet and then applying policies appropriately. Call Admission Control in Lync Server 2010 applies to both audio and video traffic, but organizations can specify different limits for each type of traffic. Both a session limit (two-way traffic) and a total limit for all sessions can be specified.

The key to successful Call Admission Control deployment is to correctly define and associate the bandwidth policy profiles to sites and links by completing the following steps:

Bandwidth Estimates

Table 1 defines the various bandwidth estimates for each protocol. Usually the typical bandwidth usage values can be used for planning purposes. Forward error correction (FEC) is enabled when Lync clients detect poor network connectivity and attempts to provide a more resilient voice connection to combat network jitter or latency.

When a bandwidth policy limit is exceeded by a user, Call Admission Control kicks in with an attempt to reroute the call. First, the call attempts to be directed over the Internet. Instead of using the WAN link, the call traverses the Internet and Edge infrastructure. If that is not possible or fails, the call can reroute across the PSTN. Whether this is allowed depends on whether the voice policy assigned to the user allows this feature. If neither Internet nor PSTN rerouting is possible, the call will attempt to be sent directly to voicemail. Lastly, if voicemail is unavailable, the call will simply fail.

It is also important to note that Call Admission Control only applies to Lync Server endpoints traversing a WAN link. Other applications transmitting data on the same WAN link are not affected by Lync Call Admission Control policies. Organizations can define a bandwidth limit for Lync traffic and still see that WAN link become saturated due to other applications. In this scenario, it makes sense to enforce QoS policies on the WAN link to ensure Lync endpoints can always place calls.

The bandwidth override policy is enforced by the receiving endpoint and not the sender. When a call is placed, the receiving endpoint leverages its subnet information and checks whether the call will exceed the bandwidth policy limit. If not, the call will be allowed.

The only clients that actually respect Call Admission Control policies are Lync 2010 endpoints. Earlier clients, such as Office Communicator 2007 R2, are not able to perform a bandwidth check when a Lync client calls. However, media calls from Office Communicator 2007 R2 to a Lync endpoint enforce Call Admission Control policies.