I have been thinking about a good strategy for keeping lot of archival reference type of data online all the time. What type of architecture would be appropriate for such purpose?
The challenge with Storage Area Network (SAN) is that it is too expensive to buy and manage especially for the information which is very infrequently accessed, need to be stored for long time and whenever access is required, the access requirement is NOW. I haven't been able to successfully make good case for establishing SAN for such purposes unless a site already has one in place.
The challenge with any type of secondary storage like tape or optical storage is the access requirement of NOW and also too messy to manage.
I am more and more getting interested in the storage brick concept utilized by TerraServer, Yahoo! and Google for hosting and managing large amount of "static" information ... looking for more details on such inexpensive implementations.