This screen graphically describes the distribution of the various
    row lengths within a particular row cache. It is necessary
    to know the distribution of the various row lengths in the
    row cache to determine how well sized the row cache is. For
    instance, knowing that each cache entry is wasting even 5 bytes
    is significant if there are 100,000 entries in the cache.

    The summary region displays three lines containing information
    about the row cache and boundary row lengths. The first line
    identifies the selected row cache name.

    The second line identifies the number of rows whose lengths
    were below the minimum selected row length (XXX-) or above the
    maximum selected row length (XXX+++). Also, the 95th percentile
    row length is displayed; this value indicates the row size that
    95% of the rows in the cache are less than or equal to.

    The third line identifies the lowest and highest row length in
    the cache. Also, the 85th percentile row length is displayed;
    this value indicates the row size that 85% of the row in the
    cache are less than or equal to.

    The information displayed is based on a scale that represents
    the row lengths to be analyzed. Initially, the screen's scale
    is based on the cache's specified row size. This scale can be
    changed using the Config menu option.

    Each vertical line of the screen identifies a row length bucket,
    based on the selected screen scale. Each asterisk (*) in the
    bucket represents a particular number of rows in the cache. Note
    that a value of 10, for instance, means that up to 10 rows are
    represented by that asterisk (*).

    The number of collection buckets is based on the display width
    of the terminal. By default, each bucket comprises 5 columns of
    the display; the number of columns can be configured by the user
    using the Config menu option. For example, setting the terminal
    width to 132 columns will allow you to display more information
    than with an 80-column terminal width. Decreasing the number of
    columns per bucket will achieve the same effect on an 80-column
    terminal width, but with a corresponding loss of precision.

    Two sliding indicators are displayed along the bottom horizontal
    axis; "A" indicates the average row size and "M" indicates the
    median row size.

    The configuration menu option allows you to fine-tune the
    distribution parameters and the display appearance. The options
    allow you more control over what row size information is
    displayed. You can:

    o  Set the number of bucket columns

       Decreasing the number of columns per bucket allows you to
       display more row sizes. Conversely, increasing the number of
       columns per bucket allows you to display fewer row sizes, but
       with more accurate size reporting (i.e. better scaling) per
       bucket. The minimum number of columns is 2 and the maximum is
       10.

    o  Set the row length range values

       Setting a specific row size range values allows you to
       determine the actual row size display range, which is by
       default based on the row cache database parameters.

    o  Eliminate/Include empty slots

    The Row Length Distribution screen is not recorded in the binary
    output file produced using the Output qualifier. Consequently,
    this screen is not available when you replay a binary file using
    the Input qualifier.