11.5 Cisco VCS and Cisco Expressway Clustering

This topic describes clustering options for the Cisco TelePresence Video Communication Server (VCS) and Cisco Expressway Series.

Cisco VCS and Cisco Expressway support clustering for scale and redundancy.

  • Up to six servers can be clustered.
  • Same OVA size must be used.
  • Cisco VCS supports a different number of servers in Cisco VCS Control and Cisco VCS Expressway clusters.
  • Cisco Expressway is limited to one remote-access domain per cluster.
  • Cisco VCS and Cisco Expressway support multiple clusters for the same domain.
  • Master replicates the configuration to all peers.

Clustering is supported. The following options are available when clustering the Cisco VCS and Cisco Expressway Series:

  • You can have up to six servers in a cluster. A cluster includes one master and up to five additional peers.
  • You must use the same OVA size for all servers within a cluster. All members of a cluster must use the same sizing option.
  • Cisco VCS supports a different number of peers in a Cisco VCS Control cluster and in a Cisco VCS Expressway cluster. For example, a Cisco VCS Control cluster with two servers can connect to a Cisco VCS Expressway cluster with four servers.
    Note

    The Cisco Expressway Series does not support different numbers of peers in the Cisco Expressway Core and Cisco Expressway Edge clusters.

  • Cisco Expressway Series remote access is limited to one domain per cluster.
  • The Cisco VCS and Cisco Expressway Series support multiple clusters for the same domain.
  • The master of each cluster replicates the configuration to all peers. All changes that are done on the master server are replicated to all other servers that are in the same cluster. If any changes are done directly on a nonmaster server, the changes are removed during the next replication.

Main Clustering Considerations

This section describes what you need to consider when implementing Cisco VCS or Cisco Expressway Series clusters.

  • The same software version must be used on all members within a cluster.
  • A DNS SRV must exist for the cluster, including A records for all members.
  • Each member must have a unique IP address.
  • The maximum round-trip delay between any two members must not exceed 30 ms.
  • NAT must not be used between any two members.
  • All members must use a platform with equivalent capabilities.
  • All members must use the same set of option keys (for high availability).
  • Each member has a unique system name.
  • H.323 mode must be enabled on all members.

Before you join servers into a cluster, you must consider the following main prerequisites:

  • All cluster members must run the same version of code. The only occasion where members are allowed to run different versions of code is for the short period of time while a cluster is being upgraded from one software version to another software version. During this upgrade period, the cluster operates in a partitioned fashion.
  • A DNS SRV must be available for the cluster, and the SRV must have an address record (IPv4 A or IPv6 AAAA) for each member of the cluster.
  • The members must have different IP addresses. The IPv4 address and the IPv6 address, where enabled, must be unique for each member.
  • Each member of a cluster must be able to reach all other cluster members within a 15-ms delay (30-ms round-trip delay).
  • Each member has a directly routable path to all other members. NAT must not be used between any two members of a cluster.
    Note

    If there is a firewall between any two cluster members, make sure that you allow IP communication between the members.

  • All members of a cluster must use a hardware platform (appliance or VM) with equivalent capabilities. For example, you can cluster servers that are running on standard appliances with servers that are running on two medium-sized VMs, but you cannot cluster a server that runs on a standard appliance with servers running on eight large-sized VMs.
  • All members must have the same set of option keys installed.
  • Each member must have a unique system name.
  • H.323 mode must be enabled on all members. Even if all endpoints in the cluster use SIP, H.323 signaling is required because it is used for endpoint location searching and for sharing bandwidth usage information between the members of a cluster.

Clustering Deployments

This section provides examples of Cisco VCS and Cisco Expressway Series clusters.

  • Same number of servers in a Cisco VCS Control or Cisco Expressway Core cluster and in a Cisco VCS Expressway or Cisco Expressway Edge cluster:
  • Unequal number of servers in a Cisco VCS Control cluster and a Cisco VCS Expressway cluster:

The figure shows two examples of Cisco VCS and Cisco Expressway clusters.

  • In the first example, the same number of servers is used in a Cisco VCS Control or Cisco Expressway Core and the attached Cisco VCS Expressway or Cisco Expressway Edge cluster. This deployment option is supported on Cisco VCS and on Cisco Expressway Series.
  • In the second example, an unequal number of servers is used in a Cisco VCS Control cluster and the attached Cisco VCS Expressway cluster. This deployment option is supported only on Cisco VCS. It is not supported on the Cisco Expressway Series. When you use the Cisco Expressway Series, you must deploy the same number of servers in the Cisco Expressway Control and the attached Cisco Expressway Edge clusters.