[Exim] Reducing Queue Sizes

Greetings Everyone-

Our organization has been running a very basic installation of Exim 3.35 for
several years now. Within the past year, our message load has increased
from about 1500 message a day to 50K+ messages a day. I have two new
servers on order, and would love to get some input on a strategy for
handling these messages more effectively. I am also planning to upgrade to
version 4.x at that time.

I currently have Exim configured to check all recipient addresses against a
list of known-good local parts. Exim then routes the messages to our
Exchange server. The problem is that I routinely see queues of 4000
messages or more resulting from spoofed source addresses (Spam). This seems
to cause the server to stop accepting SMTP connections several times a day.
Each outage seems to coincide with a scheduled queue runner process. Here
is a snippet from my configuration file:

# Route to Mail Server
punt:
driver = smartuser
local_parts = lsearch;/etc/exim/users
transport = local_smtp
fallback_hosts = <exchange.server>

local_smtp:
driver = smtp
hosts = $local_part_data

I also added the following options in hopes of solving the problem to no
avail:

smtp_accept_max = 150
smtp_accept_reserve = 40
smtp_reserve_hosts = 10.0.0.0/16
smtp_accept_queue_per_connection = 30
auto_thaw = 2d
ignore_errmsg_errors
timeout_frozen_after = 3d

Based on what I have been able to glean from the list archives, it looks
like I can configure a primary and secondary MX with a third Exim server as
a fallback host. The fallback host would handle all messages that could not
be immediately delivered. Most of the time, when the recipient address is
not located in /etc/exim/users, it is SPAM. I want to keep those messages
from clogging up my mail queue.

Are there better ways to solve this issue? I'm certain that I am not the
first one to have found myself in this situation. Any help and/or
suggestions would be greatly appreciated.

Regards,

Chad Butler