High Availability Cluster Multi-Processing (HACMP™) on Linux is the IBM tool for building Linux-based computing platforms that include more than one server and provide high availability of applications and services.
Both HACMP for AIX and HACMP for Linux versions use a common software model and present a common user interface (WebSMIT). This chapter provides an overview of HACMP on Linux and contains the following sections:
•Overview
•Cluster Terminology
•Sample Configuration with a Diagram
•Node and Network Failure Scenarios
•Where You Go from Here.

Overview:

HACMP for Linux enables your business application and its dependent resources to continue running either at its current hosting server (node) or, in case of a failure at the hosting node, at a backup node, thus providing high availability and recovery for the application.
HACMP detects component failures and automatically transfers your application to another node with little or no interruption to the application’s end users.
HACMP for Linux takes advantage of the following software components to reduce application downtime and recovery:
•Linux operating system (RHEL or SUSE ES versions)
•TCP/IP subsystem
•High Availability Cluster Multi-Processing (HACMP™) on Linux cluster management subsystem (the Cluster Manager daemon).
HACMP for Linux Cluster Overview
Overview
14 HACMP for Linux: Installation and Administration Guide
1

HACMP for Linux provides:

•High Availability for system processes, services and applications that are running under HACMP’s control. HACMP ensures continuing service and access to applications during hardware or software outages (or both), planned or unplanned, in an eight-node cluster. Nodes may have access to the data stored on shared disks over an IP-based network (although shared disks cannot be part of the HACMP for Linux cluster and are not kept highly available by HACMP).
•Protection and recovery of applications when components fail. HACMP protects your applications against node and network failures, by providing automatic recovery of applications.
If a node fails, HACMP recovers applications on a surviving node. If a network or a network interface card (adapter) fails, HACMP uses an alternate networks, an additional network interface or an IP label alias to recover the communication links and continue providing access to the data.
•WebSMIT, a web-based user interface to configure an HACMP cluster. In WebSMIT, you can configure a basic cluster with the most widely used, default settings, or configure a customized cluster while having the access to customizable tools and functions. WebSMIT lets you view your existing cluster configuration in different ways (node-centric view, or application-centric view) and provides cluster status tools.
•Easy customization of how applications are managed by HACMP. You can configure HACMP to handle applications in the way you want:
•Applications startup.You select from a set of options for how you want HACMP to start up applications on the node(s).
•Applications recovery actions that HACMP takes. If a failure occurs with an application’s resource that is monitored by HACMP, you select whether you want HACMP to recover applications on another cluster node, or stop the applications.
•HACMP’s follow-up after recovery. You select how you want HACMP to react in cases when you have restored a failed cluster component. For instance, you decide on which node HACMP should restart the application that was previously automatically stopped (or moved to another node) due to a previously detected resource failure.
•Built-in configuration, system maintenance and troubleshooting functions. HACMP has functions to help you with your daily system management tasks, such as cluster administration, automatic cluster monitoring of the application’s health, or notification upon component failures.
•Tools for creating similar clusters from an existing “sample” cluster. You can save your existing HACMP cluster configuration in a cluster snapshot file, and later recreate it in an identical cluster in a few steps.