IBM SG24-5131-00 User Manual

© Copyright IBM Corp. 1999

So, why would you prefer to install one version of HACMP instead of 
another? This chapter summarizes the differences between them, to give you 
an idea in which situation one or the other best matches your needs. The 
certification test itself does not refer to these different HACMP flavors, but it is 
useful to know the differences anyway. 

The following paragraphs are based on the assumption that you are using 
Version 4.3. For an overview of previous Versions and their corresponding 
AIX levels, as well as the supported hardware to run on, see Table 3 on page 
8. 

High Availability Cluster Multi-Processing for AIX (HACMP for AIX) Version 
4.3 comes in two flavors. One of them directly derives from previous versions, 
and therefore is called Classic, and the other, which utilizes another 
technology for heartbeating, is called HACMP Extended Scalability 
(HACMP/ES) see below for details.

Basically, these two versions differ only in the way the cluster manager keeps 
track of the status of nodes, adapters and networks. In the Classic Version, 
this is done through the use of Network Interface Modules.

Network Interface Modules  (NIMs) monitor the nodes and network 

interfaces associated with a cluster. Each 
network module monitors one cluster network 
using one kind of communication protocol (for 
example, Ethernet or FDDI). Each network 
module is responsible for maintaining 
keepalive traffic with neighboring nodes as 
directed by the Cluster Controller, for 
providing a link to other nodes on the network 
it monitors, and for initiating adapter swaps on 
certain networks.

HACMP/ES no longer used NIMs, but utilizes a technolgy that was originally 
developed on the RS/6000 SPs. Since PSSP Version 2.2. RS/6000 SP 
Systems come with the Phoenix technology for managing availability of the 
nodes. This technology was already designed as a basic instrument for