Welcome to the Monastery. You can update your post to make it more legible: put "<code>" on a line above "order_id, time_mins", and then put "</code>" on the line after the "etc...".

As for your question: it's likely that using a hash of some sort will be useful, but it's not clear what you really need. In your sample, you have two instances of "096000BP" as the "order_id", with time_mins values of "85" and "32". Do you want an output like "096000BP, 53"? What if that order_id value occurs a third (fourth...) time -- what sort of differences should be reported then?

Depending on what you really want, a simple hash might do, or you might need a hash of arrays (for each order_id value, store the list of distinct time_mins values). Something like this will load such a structure, and check for cases where a given hash key occurred multiple times in the data:

my %orders;

while (<DATA>) {
    my ( $id, $mins ) = split /[\s,]+/;
    push @{$orders{$id}}, $mins  if ( $id );
}

for my $id ( keys %orders ) {
    if ( @{$orders{$id}} > 1 ) {
        my @times = @{$orders{$id}};
        # do something with @times
    }
}

__DATA__
096000BN, 32
096000BP, 85
096000BG, 132
096000Be, 85
096000BP, 32
[download]

I don't know what you mean by "extremely large". If it's a matter of hundreds of millions of data rows, you might need some sort of "dbm" file as your hash (see AnyDBM_File), or maybe it's time to use a real database.