SIP high availability

 

SIP uses the high availability features in WebSphere Application Server to offer a comprehensive high availability (HA) solution. The SIP in WebSphere HA solution is described in the following topics:

• High availability manager

• Replicating SIP sessions

• SIP session affinity and failover

• SIP cluster routing

• Upgrading SIP applications

• SIP IP sprayer

• SIP load balancer

• Setting up a high availability environment

 

SIP HA architectural considerations

• Each container is handling state replication and SIP traffic.

• UDP/TCP/Multicast is used for state replication and SIP protocol.

• Complete session state is replicated between all servers in the replication domain.

• Each replication domain is optimal at two servers. Three servers per replication domain does not perform as well because each server must maintain a copy of all session data in the replication domain, but on a failure only half the servers replicated data is activated on any of the remaining servers.

• The number of SIP proxy servers per blade is dependent on the number of cluster addresses configured at the IP Sprayer.

 

SIP proxy configuration considerations

• No call state information is stored at the SIP proxy.

• If the IP sprayer is configured for MAC forwarding:

  • The SIP proxy must listen on the loopback address that corresponds to a cluster address configured at the IP sprayer.

  • The IP sprayer configuration dictates the number of SIP proxies per node (or blade).

  • Each configured cluster address can have at most one corresponding SIP proxy instance per node.

 

Core groups and Distribution and Consistency Services (DCS)

A core group is a set of processes which establish a high-availability domain to monitor and replicate state to one another. Each core group member communicates with one another through DCS. DCS handles elections, quorums, and heartbeats (virtual or TCP keepalive). DCS communicates with its peers via either Multicast/TCP or Reliable Multicast/UDP. The core group coordinator is elected by quorum. The default order can be changed. The core group coordinator:

• Maintains all group information – group name, group members, the policy of the group, and the state of each group member

• Assigns singleton service to group members and handles failover of singleton services.

For more information on core groups, see Core groups (high availability domains)

 

DCS failure detection

DCS failure detection includes:

• Active failure detection – a tunable heartbeat mechanism monitoring:

  • Time between heartbeats

  • Heartbeats lost before declaring failure

• TCP KEEP_ALIVE / Sockets closing

• Looking at hardware assist features

---

Sub-topics

Replicating SIP sessions

SIP session affinity and failover

Upgrading SIP applications

SIP cluster routing

SIP IP sprayer

SIP load balancer

 

Related tasks

Browse all SIP topics