Modeling your network topology

The first step towards configuring IBM® Sametime® Bandwidth Manager is to model your network topology. By understanding your network, and the locations that are connected on the network, you can identify the bandwidth bottlenecks where bandwidth constraints are wanted. In this way, you can plan the sites and links that will enable you to place constraints on bandwidth use.

About this task

The Sametime Bandwidth Manager models networks using sites and links:
  • A site represents a physical location on the corporate network, defined by a set of IP ranges. A site can represent an office location, a building, a campus, or even an entire city. The Bandwidth Manager uses the IP ranges defined for each site to associate endpoints with the site where they are located. This method assumes that IP ranges are unique within the enterprise's network.
  • The dedicated line between the two sites is modeled as a direct link between these sites. Each link is uniquely defined by the two sites it connects.

Bandwidth management for unmanaged calls

Sametime endpoints can call a SIP endpoint not registered with the Sametime SIP Registrar. Such calls are referred to as "unmanaged calls" or calls to "unmanaged endpoints."

Bandwidth Manager cannot determine the location of the called endpoint until the call is connected. When the call is connected, Bandwidth Manager can then apply bandwidth policies and constraints to future actions such as upgrading to video. The Bandwidth Manager enables an overall configuration policy for such calls that includes:
  • Whether to allow such calls to proceed even if they exceed bandwidth capacity or have no route during ICE re-INVITE.
  • The bandwidth allocation to apply to such calls

    Allowing such calls to proceed also affects NAT-traversed calls, even if they are "managed calls" in that initial IP addresses are found for both endpoints.

Bandwidth usage is configured for combinations of sites and links. Modeling your network topology helps you gain a clear overview of the sites on your network, so you can configure the allowable bandwidth and call routing values for the audio and video traffic that passes between them.

Procedure

To optimize bandwidth management, you must understand your organization's network and usage needs. For best results, model your network and settings on paper before you attempt to configure them in the Bandwidth Manager, by completing the following steps:

  1. Map your network out on paper (if possible, use an existing network map as a starting point).

    Visualizing the network with a map makes easier to identify the sites and links you should configure.

  2. Define the network's sites.

    Sites will normally correlate with buildings or cities. The simplest way to determine what should constitute a site for your enterprise is to first identify the bandwidth bottlenecks on your enterprise network. Typically, these will be places where network traffic leaves a building or campus and travels to another building or campus through a wide-area connection. These are the points where you want to control and constrain bandwidth use, so they should determine the sites you create.

  3. Define the network's links.

    A link is a connection between two sites.

  4. Define the groups to which call rates will be applied.

    A group consists of a particular class of users and endpoints (such as meeting rooms); for example, high-priority users will probably be allocated a greater amount of bandwidth during calls.

  5. Define the network's call rates.

    A call rate is a policy setting that determines the amount of bandwidth allocated for an audio or video call on a particular link. Call rates are applied to groups of users.

  6. Define the network's reflectors if they are used.

    Reflectors are used to assist in firewall and NAT traversal. Defining reflectors enables Bandwidth Manager to more accurately calculate call routes.

What to do next

Once you have a visual model of your network components plus a clear understanding of the call rates in effect and the different classes of user, you are ready to begin deployment.