Yup, hashes are great for grouping. But since you have two data points for each date (count and size), you'll need of some kind of 2d structure. I used a hash of hash in the following:

use Text::CSV_XS qw( );

my %size_by_date;

my $csv = Text::CSV_XS->new();
while ( my $row = $csv->getline($fh) ) {
    my ($name, $date, $size) = @$row;
    ++$size_by_date{$date}{count};
    $size_by_date{$date}{size} += $size;
}

die("csv parse error: " . $csv->error_diag() . "\n")
    if !$csv->eof();

for my $date (keys %size_by_date) {
    my ($count, $size) = @{ $size_by_date{$date} }{qw( count size )};
    print("$date: $count, $size\n");
}
[download]