This is kind of a kludge, and it ain't using fancy REs, but I think it does what you want.

#!/usr/bin/perl -pw

use strict;
use warnings;

our ($prev1, $prev2);

if (/DISK.*DISK/) {
    if ($prev1 =~ /cache/ && $prev2 =~ /cache/) {
        print "\n";
    }
}

if ($prev1 =~ /DISK.*DISK/) {
    print "\n";
}

$prev2 = $prev1;
$prev1 = $_;
[download]

Example execution with sample data.

me@mybox:~/sandbox
$ ./2.pl input.txt
Use of uninitialized value in pattern match (m//) at ./2.pl line 14, <
+> line 1.
cache,23778,0.0,0,0.0,0,0.0,0.0,0.00
cache,205,0.0,0,0.0,0,0.0,0.0,0.00

DISK-UTIL-PEAK-HIGH:,54.3,DISK-UTIL:,2.5

PROC-Name,PID,PYS-IO-RATE,READs,READ-RATE,WRITEs
cache,26094,0.9,1,0.9,0,0.0,0.0,0.00
cache,16046,0.0,0,0.0,0,0.0,0.0,0.00
cache,3848,0.0,0,0.0,0,0.0,0.0,0.00
cache,23778,0.0,0,0.0,0,0.0,0.0,0.00
cache,205,0.0,0,0.0,0,0.0,0.0,0.00

DISK-UTIL-PEAK-HIGH:,54.3,DISK-UTIL:,3.4  ### last line ###
me@mybox:~/sandbox
$ cat input.txt
cache,23778,0.0,0,0.0,0,0.0,0.0,0.00
cache,205,0.0,0,0.0,0,0.0,0.0,0.00
DISK-UTIL-PEAK-HIGH:,54.3,DISK-UTIL:,2.5
PROC-Name,PID,PYS-IO-RATE,READs,READ-RATE,WRITEs
cache,26094,0.9,1,0.9,0,0.0,0.0,0.00
cache,16046,0.0,0,0.0,0,0.0,0.0,0.00
cache,3848,0.0,0,0.0,0,0.0,0.0,0.00
cache,23778,0.0,0,0.0,0,0.0,0.0,0.00
cache,205,0.0,0,0.0,0,0.0,0.0,0.00
DISK-UTIL-PEAK-HIGH:,54.3,DISK-UTIL:,3.4  ### last line ###
[download]

FYI, what I think you're asking for is: if "cache" is on 2 consecutive lines followed by "DISK" twice on the 3rd consecutive line, then insert a newline before and after the "DISK" line except the last newline at the end.