This file contains one row of summary data for each successful topology split. (Terms such as “imbalance ratio” and “weight” are defined in #GUID-DC3D75A5-1C6A-4521-8174-37C3FE17BD86___topo_split_configure_58059.) The file has the following columns of information:

  • Number of the trial

  • Flags related to this trial

    This column is a string column that contains zero or more of the following strings, separated by a slash (/) if there are more than one:

    • BEST_BALANCE if this trial is the most evenly balanced trial (where “most evenly balanced” equates to the lowest imbalance ratio)

    • LOWEST_PROXIES if this trial has the lowest proxy percentage (where “proxy percentage” is the total number of proxy devices in all domains divided by the total number of proxied and owned devices in the topology split)

      Note that more than one trial can have the same flag if the trials tie for the best value on that criterion.

  • Minimum weight of any of the domains in the trial

  • Maximum weight of any of the domains in the trial

  • Average weight of the domains

  • Imbalance ratio, specified as the ratio of the maximum to average weight

    This criterion might be used in evaluating the best split; lower is better.

  • Percentage of the total number of devices across all domains that are proxies

    Another useful criterion in judging the best split; lower is better.

  • Name of the domain with the highest number of ports and interfaces

    Used with the following two fields to evaluate whether the topology split exceeds scalability estimates for the domain’s IP Availability Manager.

  • Number of ports in that domain

  • Number of interfaces in that domain


    Existing data is removed from the TSM name.trials.csv file at the beginning of each new invocation of trial learning mode.