Thanks a lot for your guidance Sir.Here is the code with latest updates,

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

my %h;
my %m;
my %db;

while(<DATA>){
  chomp;
  my @fields = split;
  my ($date,$database_name,$speed) = @fields[1,2,3];
  my ($date_hour,$minute) = split (/:/,$date) ;
  my $date_hour_minute = join (':',$date_hour,$minute) ;

  $h{$date_hour}{$database_name} += $speed;
  $m{$date_hour_minute}{$database_name} += $speed;
  $db{$database_name} = 1;
}


print "    collectionTime";
for my $db_keys (sort keys %db){
    print "    $db_keys";          # adjust spaces to line things up
}
print "\n";

for my $h_keys (sort keys %h){
    print $h_keys;                         # print the date/hour
    for my $db_keys (sort keys %db){
        print "    $h{$h_keys}{$db_keys}"; # pad with enough spaces to
+ match header
   }
   print "\n";  # this goes outside the inner loop, to end the line
}
[download]

Number of databases at anytime will be less than 10, do you think still sorting the %db hash keys once and putting them in an array will be useful here?

One more clarification, the total amount of data which needs to be computed will be around 100k records. So do you think this approach will be efficient enough? Please guide me if you think there is room for improving efficiency.