LINUX-HA: A place to start.

[Jiva DeVoe]

Jeez, overnight this list has turned from a nothing, no traffic list to
practically a high-volume list.  

I've seen alot of great ideas and points come out of alot of these
posts.  Yes, it seems like the only area of Linux that seems to need
work in HA would be service takeover (somewhat) and disk/data 
redundancy/HA.  I say somewhat for service takeover because some of the
capabilities needed for service takeover (namely IP aliasing etc) are
already in place, and can be implemented even by scripts.

Let me propose that we start by doing something that I think the
industry needs in general, defining the different *levels* of
high-availability.  Similar to how RAID has levels 0 through what? 5 or
higher?  We should begin by defining HA Levels 0 through say, 5 and the
components needed for each level.  This is something that I don't see in
the (very) well written Linux-HA FAQ, nor do I see any such standard
being used in other HA/Cluster definitions.  This leads to alot of
confusion I think.  Confusion we can clear up.  This will also allow us
to be able to say *immediately* that Linux supports Level X HA, and it
can give us a solid gameplan for moving into the higher levels.  (Since
each level would presumably be built upon the lower levels especially.)

So, anyone wanna take a first shot at this?  Throw something up?


-- 
Jiva DeVoe
MCSE
Devware Systems
mailto:jiva@devware.com