Corrupt Downloads or Packet Loss causing corrupt files (Flash, Image display, ZIP EoF Issues)

Good Morning Experts!

I seem to have collectively found myself at a loss and no search has provided me with resolution or direction.  I am a software developer using mostly .NET and am fairly familiar with Windows Server 2003 and IIS.  My question stems from months (maybe 6) of issues I have dealt with personally and issues I have dealt with on remote sessions doing software installation support.  

First of all, during remote support calls I need access to files (updates and other utilities) from my server so I have created a sub folder where I have separated my downloaded files so I can access them directly.  My issue with this process is that my server is hit or miss with downloads.  I start downloading and receive 300k download speed and then it stops like its frozen.  The connection speed then drops to 35k and just about every time will stop short as if it received the end of file packet and the File Download Manager says "Complete!".  Now when I locate the file and access it, I get a corrupt file, package or EoF depending on the type of file.  

Second, Image Display and Flash SWF Objects (and other related viewable sources) receive the same type of "packet-loss" causing the images to show half the image with gray box showing for the rest of the image or corrupt the SWF Object enough to have to clear the Temp Internet Files to remove the bad cached files.  Sometimes, clearing the temp cache is enough to fix the Flash issue but it is only a temporary fix.  It could very well happen the next time you open the browser.  This happens on remote computers (XP and Vista alike) and even on my dev computer.

Thirdly, let me explain the server.  It is working on 5 years old and is by no means a powerhouse touting 2gig Ram, a 2 cpu motherboard built from scratch before I came to work here.  I technically do not know what else lies within this beast of a server box. It is running Windows Server 2003 sp2 and currently hosts several sites through IIS, a SQL Server instance w/ multiple databases, a ColdFusion parser, a PHP environment w/ MySQL and a .NET 2.0 environment all running concurrently.   It is by far the most I have ever seen on one box but I would like to think this solution has an answer that doesnt include buying a new server which is in the works.  We have this server on a separate DMZ and our entire data center is protected by a Sidewinder Firewall in which we have had for more than 2 years now.  

And last but not least, some days the server runs like a champ.  No download issues and the sites are super fast.  I have tried several ways to isolate this issue by basically keeping my ear to the ground and watching for patterns.  We experience what I would call a small amount of traffic during the day.  Google Analytics says an average of 250 unique users to the main site.  with 5-600 page views.  Not enough to warrant bandwidth issues.  But keep in mind this is Rich content, Flash SWF Objects up to 100K files in some cases and very image driven.  

This server also provides web services to an offline application updating data (from SQL Server database) on the fly and allowing remote interaction with data on the server.  I have thought that maybe the larger data calls in the web services could be causing these packet-loss issues but I have tested and tested and no resolution.  For instance, if I am downloading a moderate file (10mb in size) and half way through a call is made to the server to dump data to a remote user, I have theorized that the server tries to regulate and serve both connections causing the straight forward download to stop, or miss a packet, or something...  all I have is a theory, no proof.  I do not know enough about network or TCP/IP protocols.

I have thought about this for so long I believe lobbying my boss for a new server may serve as simple compared to finding a resolution to my issue.  In my posting this question/issue on EE I am looking for any information on diagnosing this issue further.  I know of testing companies (ie Selenium) but do not know if this is the right site testing program.  Has anyone out there seen this or heard of this issue before.  Am I simply the end result of an overworked server?  I think I know the answer to that but I need a resolution.  This issue has taken more of my time in the past 6 months than any other issue.

In an effort to try and give any expert enough information to make a diagnosis, I have some issues in the Server Event Viewer I would like to share.  I have researched these issues when possible.  They all relate to hardware failure.

++++++++++++++++++++++++
Events Source - hpt3xx EID: 117 (detected a port timeout due to prolonged inactivity. All associated busses were reset in an effort to clear the condition)
      This is a very popular event in the System Event Viewer, it appears about every 15 seconds for approximately 20 minutes followed by
Event Source - dmio EID: 29 (Harddisk2 read error at block 12012831: status 0xc0000185)
      Followed by
Event Source - dmio EID: 22 (read error on object Disk2-01 of mirror Volume1-02 in volume Volume1 (start 12012768, length 128) corrected)
      Which seems to cause the hpt3xx event to cease the 15 second interval.  does this call fix the issue?  Because it restarts with a new hpt3xx issue.

The Event Viewer is loaded with this pattern.  Time moves further along as you go down the list below.

hpt3xx
hpt3xx
hpt3xx
hpt3xx
hpt3xx
hpt3xx
hpt3xx
dmio 29
dmio 22
(all quiet)

hpt3xx
hpt3xx
hpt3xx
hpt3xx
hpt3xx
hpt3xx
hpt3xx
dmio 29
dmio 22
(all quiet)

hpt3xx
hpt3xx
hpt3xx
hpt3xx
hpt3xx
hpt3xx
hpt3xx
dmio 29
dmio 22
(all quiet)

++++++++++++++++++++++++++++

The above is over the course of 15 minutes.   As you can see this server is a riddle wrapped in an enigma.  This type of event I have detailed here happens everyday since I began checking in on it a few months back.  I checked it about every 6 months and this was not happening this time last year.  Or if it was, I was not aware.

Thanks for reading this and taking the time to help out!

Lucas