The primary purpose of networked storage is to share it. Even in the smallest studio, you will still be sharing content with clients, senior management, or other contractors if not with other artists and developers. This occurs both locally (on the LAN) and over the Web (at the very least by email and FTP). The best technologies should help us share content, not limit its access.
The second purpose of network storage is to protect it. Depending on every user to back up and index ongoing content in development is risky at best. Given the value of content (it being the lifeblood of a studio), a centrally managed storage system provides the simplest and therefore safest way to archive and protect that investment. A system that's redundant by design protects data. And each storage technology plays an integral role in that design.
RAID provides the speed and protection for immediate production work. There's no question every network will have a RAID subsystem, but every workstation should also now having a RAID subsystem installed. The cost is nominal, and it closes the gap on potentially lost content at the point of creation.
RAID alone, however, is not enough. RAID controllers have been known to fail, corrupting an entire set and losing all the content. Also, a local RAID system can't cope with something as large as a fire on the premises, an earthquake, or a flooded data center. For these, a fast, removable storage medium is useful—enter tape. Do daily backups to recover from accidental deletions (a primary cause of lost data). Store weekly tape backups offsite in a safety deposit box or secure facility for insurance against the big stuff.
While tape does achieve the protection goal, as an archive it fails on the sharing goal. Content on tape is like suffering from amnesia. You seem to remember knowing something, but you can't recall it. Even if there's an index to tapes and such, you're still stuck with hassle of having to locate and retrieve the tape itself, then wait while you search through the data on the tape to find the content. (Oh, and do you have the appropriate drive for that type of tape?) It would seem bad policy to have the adminstrator spend hours scanning old tapes to find content. The worst scenario, though, is to pay artists to recreate that content because of the recovery hassle.
With the plummeting cost of hardware, you could keep everything on RAID. The problem here is the ever-rising cost of energy (don't forget to you have to keep all those racks of drives cool). This can and will get expensive. And this expense goes on year after year until a project comes along that requires some part of that content. Given that this could extend to 10, 15, even 20 years down the road, this surely isn't practical.
The rational solution is to use optical storage. A long-term, optical storage library is essential because it provides low-cost, yet continual access to even ancient data for both Web and internal users. It provides an ideal solution for allowing spontaneous repurposing of older content. With optical storage, it is all random access in a medium that every workstation can cope with. It achieves both our goals—it is a great way to protect and share content over the long-term.