I'm kind of partial to setting the input record separator when there is a consistent separator between records, and then it's usually possible to fashion a regex to pluck out the values from a single record:

#!/usr/bin/env perl
use Modern::Perl;

local $/ = ""; # split records on a blank line

while (<DATA>) {
    if ( m[Orderbook\ ID: \s* (.+?) \n
              \s* Symbol: \s* (.+?) \n
              \s*   ISIN: \s* (.+?) \n ]xs ){
        say "$1:$3:$2";

        # comment the previous line and uncomment the next three if yo
+u really
        # want to strip whitespace from your values, unlike your first
        # example

        # my $l = "$1:$3:$2";
        # $l =~ s/\s+//g;
        # say $l;
    }
}

__DATA__
[Sample data removed by request of the original poster.]
[download]