Lifeline uses two tiers of load balancing for workloads targeting TCP/IP applications. Lifeline directs first-tier load balancers to route workload connections to second-tier load balancers in the selected site, which then route the connections to applications in the site.
Lifeline relies on IBM MQ® clusters for workloads using messaging. Lifeline directs the cluster to route workload messages to IBM MQ queue managers in the selected site, which then make the messages available to applications.
Site routing recommendations
For workloads using two tiers of load balancers, Lifeline provides first-tier load balancers with site connection routing recommendations based on the availability and health of the workload applications, the z/OS systems and (if applicable) Linux® on IBM Z systems across both sites. For workloads using IBM MQ clusters, Lifeline provides the cluster with site message routing recommendations based on the availability and health of the IBM MQ queue managers and the z/OS systems across both sites.
Lifeline Advisors and Lifeline Agents
Lifeline consists of Advisors and Agents. An Agent is started on each z/OS system and Linux on Z Management Guest where the workloads are present, across both sites. The Agent is responsible for monitoring the workload applications that reside on its system, and to report this information back to the Advisor. The Agent on z/OS is also responsible for communicating with an MQ queue manager to monitor and influence MQ message routing within an MQ cluster.
An Advisor is started on a z/OS system and can be started as the primary or secondary Advisor. A primary Advisor communicates with all Agents to determine workload availability. The Advisor provides MQ message distribution rules to the Agents for the MQ clusters and routing recommendations to load balancers for TCP connections for these workloads. A secondary Advisor monitors the availability of the primary Advisor and will take over primary Advisor responsibility in the event of a primary Advisor failure.
Each workload that is configured to Multi-site Workload Lifeline is classified as an Active/Standby or Active/Query workload.
• An active-standby workload is active in one site. Lifeline directs load balancers to route incoming connections to the active site. When database updates are made, database software replication transmits those changes asynchronously from the active instance of the workload to the standby instance of the workload. At the standby site, the standby instance of the workload is active and ready to receive work. The updated data from the active site is applied to the database subsystem running in the standby site in near real time.
• An active-query workload can be active in one or both sites. Lifeline provides routing recommendations to the load balancers to intelligently balance connections across both sites. When database updates are made by the associated active-standby workload, database replication latency is monitored by Lifeline to ensure connections are not routed to a site if the replicated database on that site contains data that is too out of date with the database on the active site.
Lifeline can support many types of workloads that reside on z/OS or Linux on Z:
• TCP-based applications such as CICS Sockets or HTTP Servers
• SNA applications that can be accessed from a TCP-based server such as TN3270
• MQ applications that receive messages from an MQ cluster defined on z/OS
• Db2 subsystems accessed via DRDA messaging
• TCP-based applications running on Linux on Z guests such as WebSphere Application Server
IBM Multi-site Workload Lifeline V2.5 User's Guide