This is what I think of as the standard approach to this sort of problem: it's my reflexive First Thought whenever I hear or read "overlapping match":

c:\@Work\Perl>perl -wMstrict -le
"my $s = '123perl456perl789perl10';
 ;;
 my @caps = $s =~ m{ (?= (.. perl ..)) }xmsg;
 ;;
 print qq{'$_'} for @caps;
"
'23perl45'
'56perl78'
'89perl10'
[download]

It has the advantages of using only a single capture group, and of needing no look-behind, which has the limitation in Perl regex of being fixed width.

Update: Please see perlre, perlretut (especially Looking ahead and looking behind), and perlrequick.