Hi Dave,
You do have an interesting issue. I would certainly be concerned over the integrity of your data. The first thing is to preserve as much of your data as possible, I would recommend BCPing all your data out to text files and store these in a safe place.
Next you should have a look at the Page which contains the data where the write failed. Below you can use the following commands to display the page information. The offset you see is the number bytes from the start of the file, if you divide this by 8192 you will get the page number 957857.
dbcc traceon(3604) --instructs DBCC commands to display to the screen
dbcc page (dbname, 1,957857,3)
This should show the contents of the page, check to see if these are correct. Try the previous page and the next page, if these are correct, then you are in luck, your data may be intact.
You need to consult with your SAN vendor to see if they can disable the cache for your server and start investigating to see if there are any disk corruptions within the SAN on your spindles.
We do have a product that will read your backup file using the page information, so if you continue to get issues, we can assist you there, and this will certainly report back if their are any corruptions.
Regards
Douglas Chrystall
Imceda Software, Inc