If you can guarantee uniqueness of the field you're sorting on, and if you're going to be manipulating these data by the field itself, sometimes it's useful to use the sorted column as the keys of a hash and drop the rest of the data into the hash as an associative array:

#!/usr/bin/perl
use warnings;
use strict;

package main;

my %data_set = ();

while(<DATA>) {
    my @row = split/\s+/, $_;
    $data_set{$row[3]} = [ $row[0], $row[1], $row[2] ];
}

foreach ( sort { $b <=> $a } keys %data_set ) {
    printf "%-6d -> %-4d %-7s %-6d\n",
        $_, $data_set{$_}->[0], $data_set{$_}->[1], $data_set{$_}->[2]
+;
}

__DATA__
0    TOTAL    11997    26981
0    root    1089    2594
1    daemon    0    0
594    pipmp01    591    1151
958    maestro    335    685
[download]

The output is:

C:\Code>perl hashsort.pl
26981  -> 0    TOTAL   11997
2594   -> 0    root    1089
1151   -> 594  pipmp01 591
685    -> 958  maestro 335
0      -> 1    daemon  0
[download]