So your requirements are that you want to parse XML (possibly produced in-house) and that you want it to be fast,especially as you are only interested by the root tag of the document.

Would these be satisfied by using a full-fledged XML parser, but by only parsing until the first open tag?

If yes then a pull parser is what you are looking for. Just pull the XML (including potentially a DTD, silly comments and tons of PIs) until you find the tag... and then stop! XML::Parser has a pull mode, but I found that XML::TokeParser is actually much easier to use in this case:

#!/bin/perl -w
use strict;

use XML::TokeParser;

#parse from open handle
  my $p=XML::TokeParser->new(\*DATA,Noempty=>1);

#skip to <Root...>
  my $tag= $p->get_tag('Root'); 
  my $atts= $tag->[1]; # that's where the atts are, look at the docs
  
while( my( $att, $value)= each %$atts)
  { print "$att -> $value\n"; }

__DATA__
<?xml version="1.0"?>
<!DOCTYPE Root SYSTEM "/path-to/theDtd">
<Root Id="456990" Group="Navy" TimeStamp="20020116123446" Performance=
+"Regular" Database="gyt98x" Project="x">
<Request>
<DataForce>Premium</DataForce>
</Request>
</Root>
[download]

2 side notes:

• Waouh! I am impressed by the number of "use a real parser" answers to this node! Good! I can now officially retire from my position as XML Ayatollah extraordinaire ;--)
• this is the third time this week that the easiest answer to an XML question was not one of the most popular XML module (if "most popular" means SAX/DOM/XPath or even Twig). Process data from a file extracted from a relational table? XML::RAX was designed for this. Turn all elements or attribute names to upper case? XML::PYX did it in one line. I take this as a testimony to the strength of TIMTOWDI and as kind of a lesson against using only one tool for any job (and I see SAX being advocated as such a little too often for my liking). Know your modules (or ask here ;--) and live happy!