How to update network driver

I don't know Debian, but I would
- download latest driver from http://www.intel.com/support/network/sb/cs-006120.htm
- unpack
- "make install" in src directory
- "depmod -a" in case "make install" haven't done it
- reboot

if u r trying download then install later, then try some live version that supports your HW (eg ubuntu)
in other case goto system manager then remove old drivers and then try installing new one

Do not remove old drivers until you have new ones in place, are you certain it isan issue with the driver rather than the network infrastructure being saturated. I.e the switch you have does not have the necessary back plain to handle the data rate you have on the network.

well, the server has 2 gigabit nic's plugged into different switches and when running transfers etc, the disks are the bottlenecks rather than the switches.

I have been told to update the drivers to rule it out as the cause initially. Other than that I have been told to switch from Iet iscsi to another one (not sure of the name initially)

Are the network interfaces  bonded? Is the iscsi setup for multipathing or is access to the iscsi is via one interface while the other is for other purposes? One segment for the iscsi and another segment for everything else?

Do you have errors indicating the network interface shifted state?
Do you have logs on the switch for the port to which the nic is connected?? CRC errors?

No bonded interfaces, there are 2 nics on the same network, one is for iscsi to a windows server running backups, the other nic is connected to a vmware server to provide iscsi storage for this.

no errors on the switches, or anything in the logs stating that the link is going down or changing speed. Its driving me crazy, only tends to occur under heavy io loads.

Using the same segment means that the path to the VMware can bounce between the two and if your iscsi LUN allocation is limited to one nics ip, and when it switches to the other the access is denied and could explain your error.
Usually using separate segments is suggested which is how the path to the resource is guaranteed through the one path.

I had password protected each lun, and only given the password for the the esx lun to esx.

Since then I have rebuild the box with freenas 8, and all is ok now. I am having another issue with a windows ISCSI drive becoming corrupted so I will create a new question for this, thanks for your help.

I awarded 500 points when I closed the question....and the close request is still pending.

There is no info in the comment you chose to close the question. IMHO, you should pick the various comments that helped you.  You can request attention to get the question reopened.

Your comment was the only one that was helpful, hence why I awarded points to you.