Browse Prior Art Database

Method for Consistent Data Replication for Collection Management in A Dynamic Partially Connected Collection

IP.com Disclosure Number: IPCOM000102157D
Original Publication Date: 1990-Oct-01
Included in the Prior Art Database: 2005-Mar-17

Publishing Venue

IBM

Related People

Schwartz, FA: AUTHOR [+3]

Abstract

This mechanism addresses the key aspects of managing a collection where a program such as VTAM manages the overall network flows. With the use of this mechanism, APPC/VM VTAM Support (AVS) can manage the collection. This allows TSAF to be bypassed, thus increasing system performance and system reliability. This mechanism also increases the size of the collection beyond the TSAF imposed limits. This mechanism can also be used in TSAF to replace its collection management mechanisms in order to increase the collection size.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 13% of the total text.

Method for Consistent Data Replication for Collection Management in A Dynamic Partially Connected Collection

       This mechanism addresses the key aspects of managing a
collection where a program such as VTAM manages the overall network
flows.  With the use of this mechanism, APPC/VM VTAM Support (AVS)
can manage the collection.  This allows TSAF to be bypassed, thus
increasing system performance and system reliability.  This mechanism
also increases the size of the collection beyond the TSAF imposed
limits.  This mechanism can also be used in TSAF to replace its
collection management mechanisms in order to increase the collection
size.

      Transparent Services Access Facility (TSAF) currently manages
collections in VM/SP.

      A collection is a logical subdivision of a larger network of
computer processors with the following properties:
      o    Dynamic formation
           o    Systems join and leave the collection dynamically.
      o    Dynamic Resource identification
           o    Resources (Transaction Programs) can dynamically
identify themselves within the collection.
 o    Unique names (flat name space) for global resources.
 o    Global resource names are unique within the collection.

      Transparent Services Access Facility (TSAF), which is a VM/SP
component, currently manages collection formation of up to 8
processors and line flows between the processors.

      Collection management in TSAF is based on several processes:  a
join protocol, clock synchronization, atomic broadcast and collection
monitoring.

      The join protocol determines for each link-connected pair of
processors which one should join the other.

      Clock synchronization provides a basis for collection-wide
timer based functions by keeping all of the logical clocks throughout
the collection within a bounded differential of each other.  This
synchronization must be established and maintained for all nodes in
the collection in order for timeout-based broadcasts to work
correctly.

      Atomic broadcast is a method for accurately (and
simultaneously, if necessary) installing replicated data throughout
the collection.

      Collection monitoring consists of the application of timeouts
to the aforementioned process in addition to the monitoring of links.
In order to form a TSAF collection and to maintain a fully formed
collection, all joined nodes must participate accurately and in a
timely fashion relative to time parameters agreed upon via atomic
broadcast.

      A method is needed to reliably and consistently distribute data
about the members and resources of a collection of processors within
a network of computer processors in order to manage the collection
formation and activities.  With the current mechanism used in TSAF,
it is currently impossible to increase the size of a collection. They
are also not reliable as the collection approaches i...