Cisco Cisco TelePresence Video Communication Server Expressway
Appendix 5 – Configuring endpoints to work with a VCS cluster
Cisco VCS Deployment Guide: Cluster creation and maintenance (Cisco VCS X7.1)
Page 53 of 69
If the endpoint does not support DNS SRV, on startup the endpoint will perform a DNS A-record
lookup. The DNS server will have been configured to support round-robin DNS, with each of the
cluster peer members defined in the round-robin list.
lookup. The DNS server will have been configured to support round-robin DNS, with each of the
cluster peer members defined in the round-robin list.
The endpoint will take the address given by the DNS lookup and will then try and register with the
relevant cluster peer. If that peer is not available, then the endpoint will perform another DNS lookup
and will try to connect to the new VCS peer that it is given. (The DNS server will have supplied the
next cluster peer’s IP address.)
relevant cluster peer. If that peer is not available, then the endpoint will perform another DNS lookup
and will try to connect to the new VCS peer that it is given. (The DNS server will have supplied the
next cluster peer’s IP address.)
This will be repeated until the endpoint can register with a VCS. On registering with the VCS, the
Cisco will respond with the H.323 ‘Alternate Gatekeepers’ list containing the list of VCS cluster peer
members.
Cisco will respond with the H.323 ‘Alternate Gatekeepers’ list containing the list of VCS cluster peer
members.
The endpoint will continue to use the first VCS that it registered to for re-registrations and for calls. If it
ever loses connection then it will select an “Alternate Gatekeeper” from the list it was supplied with.
ever loses connection then it will select an “Alternate Gatekeeper” from the list it was supplied with.
DNS cache timeout should be set to a fairly short time (e.g. 1 minute or less) so that on failure to
reach a VCS at startup, the endpoint is quickly pointed at a different VCS.
reach a VCS at startup, the endpoint is quickly pointed at a different VCS.
Option 3 – Static IP (least preferred)
Use this option if the VCS cluster does not have a DNS name.
On each H.323 endpoint configure the Gatekeeper Settings as:
Discovery = Manual
IP Address = IP address of a VCS peer
On startup the endpoint will try and register with the VCS at the specified IP address. If that is not
available, then the endpoint will continue trying at regular intervals.
available, then the endpoint will continue trying at regular intervals.
This will be repeated until the endpoint can register with the VCS. On registering with the VCS, the
VCS will respond with the H.323 “Alternate Gatekeepers” list containing the list of VCS cluster peer
members.
VCS will respond with the H.323 “Alternate Gatekeepers” list containing the list of VCS cluster peer
members.
The endpoint will continue to use the first VCS that it registered to for re-registrations and for calls. If it
ever loses connection then it will select an “Alternate Gatekeeper” from the list it was supplied with.
ever loses connection then it will select an “Alternate Gatekeeper” from the list it was supplied with.
SIP endpoints
The options below are listed in preference order for providing resilience of connectivity of endpoints to
a cluster of VCSs where 1 or more VCS cluster peers become inaccessible. The choice of option will
depend on what functionality the endpoint you are using supports.
a cluster of VCSs where 1 or more VCS cluster peers become inaccessible. The choice of option will
depend on what functionality the endpoint you are using supports.
Option 1 – SIP Outbound (preferred)
SIP outbound allows an endpoint to be configured to register to 2 or more VCS peers simultaneously.
The benefit of this is that if the connection between one peer and the endpoint gets broken, then a
connection from the endpoint to the other peer remains. With the endpoint registering to both peers
simultaneously, there is no break in service while the endpoint realizes that its registration has failed,
before it registers to a different peer. Thus, at no time is the endpoint unreachable.
The benefit of this is that if the connection between one peer and the endpoint gets broken, then a
connection from the endpoint to the other peer remains. With the endpoint registering to both peers
simultaneously, there is no break in service while the endpoint realizes that its registration has failed,
before it registers to a different peer. Thus, at no time is the endpoint unreachable.
Configuration of SIP outbound is endpoint specific, but typically will be:
Proxy 1
•
•
Server discovery = Manual
•
Server Address =
DNS name of the VCS cluster (if DNSSRV name is available) or
DNS name of cluster peer or
IP address of cluster peer
Proxy 2