When examining your data you should probably examine what you want to take from it before deciding how to store it, in this case 2 arrays keyed on the first column best represent your intended use of the data, there's no need to create separate data structures, in fact having a single data structure makes passing it to subsequent functions easier.

So here's a solution using a single data structure with each customer referencing 2 arrays, one for time and the other for location.

~/$ cat tmp.pl
#!/usr/bin/perl
use strict;
use warnings;

open(my $customers, $ARGV[0])|| die "Couldn't open $ARGV[0] $!\n";
my %cust;
while(<$customers>){
    chomp;
    my @rec=split(/\s+/, $_);
    push @{$cust{$rec[1]}->{time}},$rec[0];
    push @{$cust{$rec[1]}->{location}},$rec[2];
}

for my $cust (sort keys %cust){
    print "$cust: ",join(" ",@{$cust{$cust}->{time}})," : ",join(" ",@
+{$cust{$cust}->{location}}),"\n";
}
~/$ perl tmp.pl tmp.dat
a: 12:10 12:13 : america australia
b: 12:11 2:30 : bombay bhutan
c: 12:12 : calcutta
n: 3:40 : neterland
[download]

Moved from duplicate thread