Not sure this is a “good” reason but it is “a” reason and one I know was cited as a concern early on.
SAN allows the server to treat it like a remote disk (i.e. you can access files by blocks and offsets as you would on a hard drive) – the NAS is a file server and typically will have its own file management “head” that takes care of this (i.e. you access files by name to the NAS and it will fetch/write the file for you) – the means you can have a much wider variety of applications/servers using a NAS without knowing or caring – SAN is going to require more work. But on the up side doing things like streaming data to/from the storage facility (i.e. recording/playing back a voice message) gives the application server using SAN much more control over how that is done. Think about control over pre-fetching chunks of a wav file for playback and controlling streaming logic on a long write – the NAS hides all this from the application, you have no control.
A real-world example: I use a NAS at home (2 TB RAID… yes, I need that much space. Really). I can stream 6 or 7 HD TV shows off the NAS to computers/TVs around the house no sweat. However recording HD TO the NAS directly is a disaster – massive file fragmentation even after I fiddled endlessly with various settings and it can’t manage more than 2 at a time without dropping frames. I ended up having to record all streams locally and then copy them over to the NAS as whole files to keep things sane and tidy. If I were using a SAN and had a media capture setup sophisticated enough to use it properly this wouldn’t be a problem.
Understood NAS and SAN are starting to blur as time moves ahead but this is certainly a viable issue to take into consideration.
