In a single head environment all of NVram is utilized to log write operations.
NVRAM is split into two pages - p0 and p1.
In an HA Pair the NVRAM on each member is split, and all writes are logged
on the local and partner NVRAM, for both p0 and p1.
NVRAM is good for about three days. I've got a real interesting story about
NVRAM and cluster failover involving hurrican Ike, but maybe I'll save that
story for later.
I hope this response has been helpful to you.
At your service,
Eugene E. Kashpureff
ekashp@kashpureff.org
Senior Systems Architect / NetApp Certified Instructor
http://www.linkedin.com/in/eugenekashpureff
(P.S. I appreciate points for helpful or correct answers.)