This chapter provides information about the clustering feature
of
Cisco Unified Communications Manager which provides a mechanism for distributing
call processing and database replication amongst multiple
Cisco Unified Communications Manager servers that run the exact same version of
Cisco Unified Communications Manager. Clustering provides transparent sharing of
resources and features and enables system scalability.
This topic provides an overview of the steps that are
required to install and configure a
Cisco Unified Communications Manager cluster, which comprises a set of
Cisco Unified Communications Manager servers that share the same database and
resources.
Procedure
Step 1
Gather the information that you need to install
Cisco Unified Communications Manager and any other software applications on the
first node and subsequent servers. Also, determine how you will allocate the
servers in the cluster.
Step 2
Install the database server (first node).
See the installation documentation for the hardware components
that you are installing.
Step 3
Install
Cisco Unified Communications Manager and any additional software applications on
the subsequent servers.
Note
Before installing the subsequent servers, you must define the
nodes in the Server Configuration window in
Cisco Unified Communications Manager Administration.
Step 4
Configure device pools and use them to assign specific devices to
a
Cisco Unified Communications Manager group.
Step 5
If you are using an intercluster trunk, install and configure it
as an intercluster trunk, either gatekeeper-controlled or
non-gatekeeper-controlled.
Step 6
If you want to provide call admission control for an intercluster
trunk, configure either a gatekeeper-controlled intercluster trunk or
Cisco Unified Communications Manager locations.
Clusters
A cluster comprises a set of
Cisco Unified Communications Manager servers that share the same database and
resources. You can configure the servers in a cluster in various ways to
perform the following functions:
Database replication
TFTP server
Application software server
You can use various nodes in the cluster for call-processing
redundancy and for load balancing.
You can activate feature services on various nodes in the
cluster to specify which servers perform certain functions for the cluster. By
accessing the Service Activation window in
Cisco Unified Serviceability, you can dedicate a particular server to one function
or combine several functions on one server, depending on the size of your
system and the level of redundancy that you want.
Tip
The Restart Cisco Communications Manager on Initialization
Exception service parameter determines whether the Cisco CallManager service
restarts if an error occurs during initialization. This parameter defaults to
TRUE and, with this value, the Cisco Communications Manager initialization
aborts when an error occurs during initialization. Setting the value to FALSE
allows initialization to continue when an error is encountered. You can locate
this clusterwide parameter in the
Database replication in a cluster
A cluster comprises a set of
Cisco Unified Communications Managers servers that share a common database. When
you install and configure
Cisco Unified Communications Manager, you specify which servers belong to the same
cluster. A cluster comprises the first node (publisher) and subsequent nodes
(subscribers). The first node in a cluster contains the
Cisco Unified Communications Manager database, which gets automatically installed
when you install
Cisco Unified Communications Manager on the first node.
Cisco Unified Communications Manager uses all subsequent nodes in the cluster for
database replication. After you add the subsequent node to the Server
Configuration window in
Cisco Unified Communications Manager Administration and install
Cisco Unified Communications Manager on the subsequent node, the node contains a
replicate of the database that exists on the first node.
After you add, update, or delete configuration in
Cisco Unified Communications Manager Administration,
Cisco Unified Serviceability, or
Cisco Unified CM User Options,
Cisco Unified Communications Manager writes the configuration update to the
Cisco Unified Communications Manager database on the first node in the cluster and
then updates the database replicates on the subsequent nodes. If both the first
node and subsequent nodes are available, you read and write configuration data
in the GUIs on the first node, even when you browse to GUIs on the subsequent
node(s) in the cluster. If the first node is unavailable, you can read
configuration data in the GUIs on the subsequent node(s), but you cannot make
updates in the GUIs on the subsequent nodes.
Consider the following information that is related to
Cisco Unified Communications Manager database replication:
Before you install
Cisco Unified Communications Manager on the subsequent node, you must add the
subsequent node to the Server Configuration window by accessing
Cisco Unified Communications Manager Administration on the first node. For more
information on adding a subsequent node to
Cisco Unified Communications Manager Administration, see
Balanced call processing.
For
Cisco Unified Communications Manager database replication to occur, you must
install the exact same version of
Cisco Unified Communications Manager on the first node and subsequent node(s) in
the cluster.
Do not make configuration changes (additions, updates, or
deletions) during a
Cisco Unified Communications Manager upgrade. If you make configuration changes
during an upgrade, you may cause data to be lost or cause data not to
replicate; in addition, the upgrade may fail.
You can view the Unified CM Cluster Overview report in
Cisco Unified Reporting to determine how all nodes are classified in the database;
that is, if the node serves as the first (publisher) or a subsequent
(subscriber) node. Likewise, you can click the Host Name/IP Address link in the
Find and List Servers window in
Cisco Unified Communications Manager Administration; after the Server Configuration
window displays, you can view the read-only Database Replication field. If the
field displays Publisher, the node serves as the first node. If the field
displays Subscriber, the node serves as a subsequent node.
Changing the name or IP address of a node in a cluster impacts
Cisco Unified Communications Manager database replication. Before you change the
name or IP address of a node, review the document, Changing the IP Address and
Host Name for
Cisco Unified Communications Manager Release 8.5(1).
To verify the state of
Cisco Unified Communications Manager database replication, for example, whether
replication is occurring, broken, and so on, you can use the
Real-Time Monitoring Tool,
Cisco Unified Reporting, or the Command Line Interface (CLI).
If you determine that a problem exists with
Cisco Unified Communications Manager database replication, you can repair database
replication via the Command Line Interface (CLI).
If you revert to a previous version of
Cisco Unified Communications Manager, you must reset
Cisco Unified Communications Manager database replication via the Command Line
Interface (CLI) after you revert to the previous version.
Intercluster communication
In very large environments, you might need to configure more than one cluster to handle the call-processing load. Communication between the clusters typically occurs by means of intercluster trunks or gatekeeper trunks. Most large systems use one of two main types of multicluster configurations:
Large, single campus, or metropolitan-area network (MAN)
Multisite WAN with distributed call processing (one or more Cisco Unified Communications Managers at each site)
Because intercluster trunks in a MAN usually have sufficient bandwidth, they do not require any call admission control mechanism. Multisite WANs with distributed call processing typically use gatekeeper technology for call admission control.
Intracluster Communication
Cisco Unified Communications Manager also supports intracluster communication, which is a multisite WAN with centralized call processing (no Cisco Unified Communications Manager at the remote site or sites). Multisite WANs with centralized call processing use the locations feature in Cisco Unified Communications Manager to implement call admission control.
Most features of Cisco Unified Communications Manager do not extend beyond a single cluster, but the following features do exist between clusters:
Basic call setup
G.711 and G.729 calls
Multiparty conference
Call hold
Call transfer
Call park
Calling line ID
For more information about intercluster communication and call admission control, see Cisco Unified Communications Solution Reference Network Design (SRND).
Balanced call processing
After installing the
Cisco Unified Communications Managers that form a cluster, you should, as much as
possible, evenly balance the call-processing load across the system by
distributing the devices (such as phones, gateways, CTI route points, CTI
ports, and route lists) among the various
Cisco Unified Communications Managers in the cluster. To distribute the devices,
you configure
Cisco Unified Communications Manager groups and device pools and then assign the
devices to the device pools in a way that achieves the balance that you want.
Cisco Unified Communications Manager groups and device pools represent
logical groupings of devices that you can arrange in any way that you want. For
ease of administration, make sure that all the devices in a group or pool share
a common and easily identified characteristic, such as their physical location
on the network.
You can also use
Cisco Unified Communications Manager groups to establish redundancy (backup call
processors) for the primary
Cisco Unified Communications Manager in the group. A
Cisco Unified Communications Manager group comprises an ordered list of up to three
Cisco Unified Communications Manager servers. During normal operation, the first
(primary)
Cisco Unified Communications Manager in the group controls all device pools and
devices that are assigned to that group. If the primary
Cisco Unified Communications Manager in a group fails, control of the device pools
and devices that are registered with the primary
Cisco Unified Communications Manager transfers to the next
Cisco Unified Communications Manager in the group list.
For example, assume a simplified system that comprises three
Cisco Unified Communications Managers in a cluster, with 300 existing
Cisco Unified IP Phones and provisions to auto-register new phones as they are added
later.
The configuration includes four
Cisco Unified Communications Manager groups: group G1 that is assigned to device
pool DP1, group G2 that is assigned to device pool DP2, group G3 that is
assigned to device pool DP3, and group G4 that is assigned to device pool DP4.
Group G4 serves as the default group for devices that auto-register.
Unified CM1 serves as the primary
Cisco Unified Communications Manager for the devices in DP1 and DP2, first backup
for DP3, and second backup for the devices in DP4.
Unified CM2 serves as the primary
Cisco Unified Communications Manager for the devices in DP3 and DP2, first backup
for DP1, and second backup for the devices in DP2.
Unified CM3 serves as the first backup
Cisco Unified Communications Manager for the devices in DP2 and DP4 and second
backup for the devices in DP1 and DP3.