The following sections provide information about Enhanced Location Call Admission Control architecture.
Model-Based Call Admission Control
Enhanced Location Call Admission Control (CAC) is a model-based CAC mechanism. The administrator creates a model of the network and how the network infrastructure handles the media.
The more accurate and detailed the model of the network is, the more effective the management of the bandwidth and avoidance of congestion is within the network. However, the model cannot account for transient network failure conditions.
Through the Cisco Unified Communications Manager interface the administrator configures the Enhanced Location CAC mechanism based on the network model.
After the administrator creates the model and enters it into Cisco Unified Communications Manager database, Location Bandwidth Manager (LBM) calculates the effective paths between all originating and terminating locations, and deducts bandwidth from each link and location along that path.
When a call is admitted between two locations, LBM deducts (reserves) bandwidth from each link and location along that path for the duration of the call. The bandwidth deduction is symmetric (bidirectional). For example, for a G.711 audio call, 80 kb bandwidth is deducted from the audio allocation assigned to each link and location in the call path. When a call is terminated, LBM restores the bandwidth deduction.
The administrator may assign bandwidth allocations to locations as well as to links, if it is desired to limit admission of intra-location as well as inter-location calls.
The intra-location bandwidth allocations are unlimited by default.
Location Bandwidth Manager
A Location Bandwidth Manager (LBM) can reside and run on every Cisco Unified Communications Manager node, or on a few selected Cisco Unified Communications Manager nodes within the cluster. LBM is a feature service and can be started and stopped from the serviceability configuration page.
Main functions of Location Bandwidth Manager are:
- Model Formation and path determination
- Replication of the model to other LBMs within the cluster, and between clusters
- Servicing bandwidth requests from Unified CM
- Replication of bandwidth deductions to other LBMs within the cluster, and between clusters
- Provide configured and dynamic information on request to Serviceability
- Update Location RTMT counters
When LBM service is started, it reads configured location information from the local database. This includes configured locations; audio, video, and immersive video capacities in those locations; links from a given location to other locations, the weight associated with those link; and the audio, video, and immersive video capacities on those links. It creates a local model with these values. Other LBMs in the cluster have access to the same data from the database and thus create the same local model at their startup. The LBM is now synchronized with the rest of the cluster and is ready to provide service.
Each Cisco Callmanager service communicates with LBM services within the cluster, as designated by an LBM group. By default, each Cisco Callmanager service communicates with the local LBM within the cluster.
Each LBM service communicates with all other LBMs within the cluster and may communicate through LBM Hubs with LBM services in other clusters. LBM services within the cluster are fully meshed.
The LBM service computes the effective path from the source location to the destination location by adding the weight of each link for each possible path between source and destination. The path with the least cumulative weight is designated as the effective path. If there is more than one path that has the same weight LBM chooses which path to use. All calls that have the same source and destination locations use the same path.
The following figure provides an example, demonstrating the calculation of the effective path from Hub_none to Loc_14:
Figure 3. Location CAC Effective Reservation Path Determination
- A path from Hub_none through Loc_12 to Loc_14 is the effective path with a total weight of 20.
- A path from Hub_none to Loc_14 has weight of 60 which is greater than 20 and therefore not the effective path.
The following are some important considerations:
- LBM group configuration allows the administrator to select the LBM service that Cisco Unified Communications Manager can communicate with.
- It is not necessary to run the LBM service on every Cisco Unified Communications Manager node.
- The administrator can configure the LBM group based on consideration to minimize the network delay for bandwidth deduction.
- The LBM group can provide redundancy of LBM service to maintain the availability of CAC mechanism during network outage.
- When Cisco Unified Communications Manager is trying to locate the LBM service to communicate with:
- It honors the LBM group association if one exists
- If there is no LBM group assigned or an empty LBM group is assigned, Cisco Unified Communications Manager uses a local LBM if it is activated
- If there is no LBM available, then Cisco Unified Communications Manager uses a service parameter to determine how to treat the call
When selectively activating LBM services and configuring the LBM groups consider the following:
- Activate at least one LBM on each distinct call processing site. Consider activating LBM on standalone servers.
- For split data center deployment, activate at least one LBM for each data center.
- Consider activating LBM on the stand-by servers where there are active and stand-by servers to reduce the impact on the active servers.
- Connect to a local LBM service when available.
- For clusters with multiple sites, select LBM services in the data center or in the closest regional site.
Inter-Cluster Location Call Admission Control
With the model-based Location CAC between clusters (intercluster), each Cisco Unified Communications Manager cluster has a local model that it controls. Through an intersystem replication mechanism, each system in the enterprise network propagates its local model to other systems and creates a global model of the entire enterprise network by putting in each model from the remote systems, and storing it in the internal memory. LBM services in each system in the enterprise network that participate in the intercluster location CAC, has the global model stored in its local memory.
When a call is made across the clusters, originating and terminating systems pass their locations and call identifiers to each other through the signaling protocol (e.g. SIP signaling protocol). Terminating and originating clusters reserve location CAC bandwidth end to end locally, using its global location CAC model, and then replicate the bandwidth reservation to other systems in the enterprise network.
The amount of intersystem bandwidth replication messages can be significant. Select LBM Hubs carefully to make replication more efficient within the enterprise network.
Race conditions may occur as each local system reserves bandwidth from the global model and then replicates the deduction. When race conditions occur, calls may be admitted in excess of those for which bandwidth is deducted.
When modeling the network, use conservative bandwidth capacity assumptions to allow for the fact that calls may be admitted in excess of those for which bandwidth is deducted.
Intercluster Location Call Admission Call Configuration Considerations
The following are some considerations when configuring intercluster location CAC between a local cluster and a remote cluster:
- The local administrator must configure the remote locations adjacent to local locations and the links between local and remote locations.
- When the local cluster receives a model replication from a remote cluster, it joins the models by identifying locations and links that appear in both models and forms a global network model.
- It is critical to name locations consistently in all clusters, to ensure the global network model assembles correctly. Follow the principle of same location, same name; different location, different name.
If there is a conflict in bandwidth capacity or weight assignment on the common links or locations, the local cluster uses the minimum of the assigned values.
Intercluster Location Call Admission Control Replication
An Enhanced Location CAC LBM replication network is used to replicate the model topology, and bandwidth deduction across multiple clusters, and within the cluster. All LBM services are fully connected within the cluster and all LBM Hubs are fully connected between clusters. LBM services that are not LBM Hubs participate in intercluster replication only through the LBM Hubs in their cluster.
The LBM Hub Group provides the mechanism for an LBM Hub to find out how to communicate with other LBM Hubs in remote clusters. By this mechanism, the LBM Hub builds a fully meshed replication network with all other LBM Hubs.
Location Bandwidth Manager Hubs
The following describes Location Bandwidth Manager (LBM) Hubs:
- An LBM service becomes a Hub when an LBM Hub Group is assigned to it.
- If a cluster has multiple LBM Hubs, the LBM Hub with the lowest IP Address functions as the sender of messages to other remote clusters.
- The LBM Hub organizes its links to remote LBM Hubs by the ClusterId assigned to it.
- The LBM Hub that functions as the sender for messages, and picks the first LBM Hub of each cluster to send messages to.
- The LBM Hub that receives messages from the remote clusters, forwards the received messages to other LBM services within the cluster.