RFC 2155 (rfc2155) - Page 3 of 124
Definitions of Managed Objects for APPN using SMIv2
Alternative Format: Original Text Document
RFC 2155 Definitions of Managed Objects for APPN June 1997
The directory and routing functions enable an NN to find where an LU
is located in the network, and calculate the optimal route for the
session based on the requested class of service (COS). A network
node saves the LU information in a directory database, which is built
from LUs defined locally, LU registration from served end nodes, and
LUs learned from network searches.
Each NN maintains a local COS database that assigns a routing weight,
or relative cost, to each resource for each class of service. For
example, the #INTER COS assigns a lower weight to TGs with a greater
effective capacity, while the #BATCH COS favors TGs with a lower
relative cost per byte.
A node saves network topology information (on NNs, VRNs, and TGs
between them) in a network topology database. The topology
information includes state and routing characteristics. Topology
information is exchanged between NNs over CP-CP sessions such that
the database is fully replicated at each NN. Information on TGs from
NNs to ENs are kept in a local topology database. Local topology
information is shared with other NNs only during the session
establishment process, to give the NN responsible for route
calculation the necessary information for end-to- end route
calculation.
SNA names such as LU names, CP names, COS names, and mode names can
be padded with blanks (space characters) in SNA formats. These
blanks are nonsignificant. For example, in a BIND Request Unit (RU)
a COS name of "#INTER" with a length of 6 is identical to a COS name
of "#INTER " with a length of 8. However, in this MIB,
nonsignificant blanks are not included by the agent. Using the COS
name from the previous example, an agent would return a length of 6
and the string "#INTER" with no blanks for appnCosName, regardless of
how it appears in the BIND RU or in internal storage. The lone
exception is the all blank mode name, for which the agent returns a
length of 8 and the string " " (8 blank spaces). The MIB
variables that this applies to are identified by a textual convention
syntax that also describes this behavior.
When an SNA name is functioning as a table index, an agent treats
trailing blanks as significant. If a management station requests the
objects from a row with index "#INTER ", the agent does not match
this to the row with index "#INTER". Since an agent has no
nonsignificant blanks in any of its table indices, the only reason
for a Management Station to include them would be to start GetNext
processing at a chosen point in a table. For example, a GetNext
request with index "M " would start retrieval from a table at
the first row with an 8-character index beginning with "M" or a
letter after "M".
Clouston & Moore Standards Track