DNS and Routing issue

Topology: 1 headquarters with 2 branch offices connected through VPN Site to
Site tunnels and 1 branch office with a dedicated leased line and connected
with a router. 1 vpn branch office has no DC and uses the Headquarters for
authentication and DNS Queries. The other two have their own DC's and Local
DNS Servers. all DC's are windows 2003

History: We have just installed a SAN. The recommended procedure is to
install 2 extra nic cards in the servers that will use the SAN. 1 is for the
active and 1 is for passive or redundancy in the event the active fails. We
then created a separate VLan and subnet for this communication to take place.
Our business is 10.10.x.x /16 and the SAN VLan is 10.15.x.x /16

We have three gateways....1 is for connecting the branch offices and
external traffic. 1 is for connecting the dedicated branch office. and an ISA
box that routes traffic between the business network and the SAN VLan for
administrative puposes only. apropriate routes have been set up and all
communications between branches and headquarters works well.

Problem: The problem I am having is with our exchange server. We have two
servers that connect to the SAN for storage. 1 is our file server, and the
other is the exchange server for mailbox stores. Both servers have 2 extra
nics installed for the SAN configuration. and Both Nics are set up identical.
But for some reason, we keep losing a connection to the exchange server.

Trouble shooting: I have already tried some things. Doing reasearch I
configured the binding order for the Business Nic to the top of the order
through Network Properties/advanced/advanced settings

I also did some research and configured the DNS Settings for the SAN VLan
Nic to not register this connections addresses in DNS under properties of
TCP/IP/advanced/DNS.
We also have WINS. So I disabled NetBios over TCP/IP so the SAN VLan nics do
not register with wins However with all the above, the behaviour is still the
same

Behavour: For testing and troubleshooting I do a continuous Ping to the
branch subnet that will suddenly stop communicating. As soon as I start to
get Request Timed out, I will open another cmd session and do a tracert
command. The results are that it is trying to go through the SAN_VLAN subnet
to the destination, and since I do not have a static route set on that
branches router, it will obviously fail. I will then do an ipconfig /register
DNS, and as soon as the command completes, the pings start to reply
immediately. I then do a tracert command and now it is going through the
business subnet the way it should. but after about 5 minutes, it will
suddenly stop again. what is even more odd, is if I just do nothing, it will
correct itself after about a minute. but it keeps toggling back and forth

So My question is??

Can anybody shed some light as to why this behavour is happening?? Where
should I look to now?? and has anybody else ran into this problem?? I am at a
loss??

Thanks

Rayzor