Lifeline Advisors

For a workload consisting of server applications, the primary Lifeline Advisor provides recommendations that are used by a first-tier external load balancer to determine how to route new requests to a second-tier load balancer to select a target site. The second-tier load balancer then routes each request to a target z/OS® system in the site to select a target application.

Rule: In this information, whenever the term Lifeline Advisor or Advisor is mentioned, it is referring to the primary Lifeline Advisor. Any references to the secondary Lifeline Advisor are indicated by the terminology secondary Lifeline Advisor or secondary Advisor.
The following are the supported types of second-tier load balancers:
  • The second-tier load balancer can be an external load balancer. It uses the Lifeline Advisor recommendations to determine how to route these new requests to the target applications in the z/OS site.
  • The second-tier load balancer can be an intermediary node. It uses its own recommendations to determine how to route these new requests to the target applications in the z/OS site.
  • The second-tier load balancer can be a z/OS sysplex distributor. It creates its own recommendations that are similar to the ones that the Lifeline Advisor creates to determine how to route these new requests to the target applications in the z/OS site.

The Lifeline Advisor maintains at least two, and up to four, listening sockets or ports; one is for the Lifeline Agents to connect to, up to two are for the external load balancers to connect to, and optionally one is for the secondary Lifeline Advisor to connect to. There are separate IPv4 and IPv6 listening sockets for external load balancers. If your TCP/IP stack is not IPv6 enabled, you will not be able to use the IPv6 listening socket.

The Lifeline Advisor maintains an access control list that allows external load balancers, the secondary Lifeline Advisor, and the Lifeline Agents to connect to it. The access control list specifies the remote IP address of the connecting external load balancers, the remote IP address and port of the secondary Lifeline Advisor, and the Lifeline Agents that are allowed to connect to the Lifeline Advisor.

For a workload consisting of MQ cluster queues, the primary Lifeline Advisor provides recommendations that are used by Lifeline Agents to control the site and target systems where MQ cluster queues are enabled for messages. Lifeline Agents interact with MQ queue managers to prevent the inactive site from receiving messages.

Recommendations provided by the Lifeline Advisor

The recommendations provided by the Lifeline Advisor are dynamic, and can change as the conditions of the target sites, systems, and server applications or MQ cluster queues change. The recommendations include the following information:
  • An indication of whether the target server application or MQ cluster queue, the target system, or the target site is currently active. The Advisor uses this information to exclude systems or sites that are not active or do not have the required server applications running or MQ cluster queue available
  • For target server applications or MQ cluster queues that reside on z/OS, z/OS Workload Management (WLM) server-specific recommendations. WLM recommendations provide a relative measure of the ability of a target system to handle new workload, as compared to other target systems across both sites. They also provide a measure of the performance of individual server applications or MQ queue managers that host the MQ cluster queue, as compared to the WLM policy goals specified for that workload. The WLM recommendations are derived from the following metrics that are based on the importance of the workload, as defined by the WLM policy:
    • Displaceable general CPU capacity of each system
    • System z® Application Assist Processor (zAAP) capacity
    • System z Integrated Information Processor (zIIP) capacity
    • The proportion of each type of CPU that the server application is using
  • For target TCP server applications that reside on z/OS, an indication of server application health from a TCP/IP perspective. TCP/IP statistics for target applications are monitored to determine whether specific server applications are encountering problems that prevent the applications from keeping up with the current workload. For example, is a target TCP server application keeping up with TCP connection requests? Are requests being rejected because the backlog queue is full? In such scenarios, the recommendations passed back to the external load balancers are adjusted appropriately, so that the load balancer can direct fewer connections to any server application that is experiencing these problems. These recommendations are provided for TCP server applications and are referred to as Communications Server weights in this information.