Providing Open Architecture High Availability Solutions

Providing Open Architecture High Availability Solutions
99
The management middleware also controls service availability groups (the collections of managed
components in redundancy relationships such as 2N or N+1) and makes role assignments (active,
standby, spare) within them.
10.7 Providing Administrative Access and Control
The availability functions of the management middleware described so far comprise an automatic,
self-managing system. The administrative functions described here provide the necessary link to
human and enterprise management system intervention and control. The goal of high availability is
greatly supported by simplified management, including a single point of access to a unified,
system-wide representation.
Effective external administration requires local and remote access to the managed components—a
central location for configuration and management of an entire cluster, the service availability
groups, and individual managed components. Administrators must be able to monitor the status of
managed components, configure the components, and receive alerts when errors occur in the
cluster. Administrators need both system-wide views and drill-downs to individual components.
Management middleware must provide a set of services for the ongoing maintenance and
management of the system. Outages due to planned system maintenance or hardware or software
upgrades are not acceptable in many high availability system designs. Among the important
functional requirements are remote, automatic upgrading; retrieval of components data;
provisioning and deploying additional nodes; integration into CIM management structures,
notification of alert conditions, and integration with SNMP network management systems.
10.8 Providing a Consistent Interface to Applications
In many HA system designs, application developers must incorporate capabilities into the
application to interface with the management middleware. There is a strong desire to keep
interfaces consistent so that applications are portable across various implementations of platforms,
operating systems and management middleware. These interfaces are discussed in detail in
Section 5.4.
10.9 Alignment with Standards
Much work has been done throughout the industry to develop standards in the area of system
communications and management. Where possible, the management middleware should embrace
these standards. Among other things, this enables system manufacturers to preserve previously-
developed or newly-designed capabilities within other parts the system. Examples of relevant
standards include:
• CIM – Common information model
• CORBA – Common Object Request Broker Architecture
• CMIP – Common Management Information Protocol
• SNMP – Simple Network Management protocol
• X.731 – ITU recommendation State Management
References for these standards can be found in the bibliography.