Active/Active with separate storage or with shared storage?

Am wondering about the new Active/Active functionality as it relates to how the systems would be configured.

Some other active/active solutions have two servers/heads/controllers which are connected to an shared external storage (Nexenta for example). Write operations sent from 1 system to the other, but only 1 of the system actually performs the write (the other 'monitors' to ensure that the write completes -- if not if takes over and completes the operation). So, in this case, the write is considered completed as soon as the other server acknowledge the pending operation -- the actual disk operation can take place later.

Others solutions, each server has it own storage and disk writes are considered complete when both servers have performed the write (disk controller caching is allowed).

Which if these models does v7 implement/follow?


Personally, I think that the first approach is better since the write operations will be considered complete more quickly (I could have a SSD based primary and 5400 RPM SATA secondary).