[cgl_discussion] USE CASE - Availability Management Framework

Steven Dake sdake at mvista.com
Mon Apr 18 12:27:47 PDT 2005


This has been published to the usecase repository for a few weeks but
there have been no comments so far.  I have included the text version
for your review.

Thanks

Description
Many applications have availability objectives that cannot be met with
standard systems solutions. There are two approaches to increase
availability. The MTBF can be very large for small values of MTTR. The
remaining approach is to reduce the MTTR to a very small value.

The Availabilty Management Framework reduces MTTR by adding redundant
software and hardware components to a system design. Applications are
divded into components which register with the AMF. These components are
then directed to either active or standby service. When an active
component fails for some reason, the standby component transistions to
active and provides the appropriate services. 


Desired Outcome
A standard interface for application failover increases availability of
Linux applications. By providing a standard interface, the developer
community with availability concerns can address these issues with
reduced effort and cost.

Currently there is one open source project that implements Availability
Management Framework. The openais project has this implementation, but
it is based upon the A.01.01 specifications. The CGL working group
desires a B.01.01 open source implementation.

Several companies are developing proprietary implementions of the AMF
for the B.01.01 specification, but these implementations are beyond the
scope of the working group.

An OSDL Special Interest Group (SIG) has been established for ongoing
discussions regarding common open source clustering services. It is
likely that this group will define the common cluster services and drive
implementations into the kernel (where needed). See
http://developer.osdl.org/dev/clusters/. Andrew Morton has stated
publically that common clustering services will need to be supported by
the clustering community and not just single clustering projects.


Participants/Roles
      * AMF Developers: Both open source and proprietary cluster
        implementations are being developed with SA Forum interfaces.
        
      * Application developers: Applications which require increased
        availability by reducint MTTR are developed to the SA Forum AMF
        interface specification.
        
        
Applications and services that can benefit from using a the availability
management framework are services for which interrupted service is
unacceptable.

Implementation Notes
The AMF should maintain correct state at all times. When a network
partitions or merges, every processor should contain a consistent view
of active and standby components.

It is desireable although not required for management of components to
be made available. It would be helpful to be able to lock or unlock a
component or service group such that the administrator may control
components in a running system. No published AMF implementation has this
functionality.





More information about the cgl_discussion mailing list