Highly Available HP-UX Internet Services (May 2008)

Executive Summary
This white paper discusses how to configure HP-UX Internet Services components:
DNS, FTP, and Sendmail, into a Multi-Computer (MC)/ServiceGuard package.
Intended Audience
This white paper is intended for HP customers who are using or planning to use DNS,
Sendmail, or FTP in an HP ServiceGuard environment. This white paper is also intended
for system administrators, HP support personnel for DNS, FTP, and Sendmail, HP field
engineers, and consultants who advise customers on High availability solutions. Readers
of this document must be familiar with using DNS, FTP, and Sendmail.
Introduction
High availability (HA) is a computing environment that is becoming increasing popular
in organizations that demand zero downtime or minimum failover time. HP
ServiceGuard is HP's high availability product that enables you to configure multiple
computers (nodes) into a highly available "cluster". It supports multiple applications
in a cluster of up to eight nodes. HP ServiceGuard monitors the health of each node
and quickly responds to failures in a way that eliminates or minimizes application
downtime. HP ServiceGuard automatically detects and responds to failures in the
following components:
• System processors
• System memory
• LAN media and adapters
• System processes
• Application processes
In a HP ServiceGuard environment, all systems in the cluster are active peers, wherein
each system can serve one or more applications. When a system fails, the application
is automatically moved to another system in the cluster configured to run that
application. The Internet Protocol (IP) address associated with that application is moved
to the adoptive node so that clients can connect to the same IP name or address; this is
a “move the service point” paradigm. Because the adoptive system does not reboot, a
node need not stop serving its own applications except for performance reasons.
HP ServiceGuard not only monitors the health of the nodes in the cluster and the health
of the application, it also monitors the health of network host adapters and switches
to a redundant adapter depending on the availability. In case of an application failure,
the application can be restarted or moved to another system in the cluster. HP
ServiceGuard ensures exclusive access to the disks belonging to an application to
prevent inadvertent access by another system connected to the same disk bus.
Executive Summary 5