R: Network mirroring (was:R: Questions)

[Marcelo Tosatti]

On Fri, 5 Nov 1999, Giampaolo Tomassoni wrote:

> > On Wed, Nov 03, 1999 at 02:48:02AM -0200, Marcelo Tosatti wrote:
> > > I know someone at NASA who wrote a GNBD kernel server but i didnt find the
> > > code. Anyone has it?
> >
> > I think someone wrote a NBD kernel server. I am not aware of a kernel
> > server for gnbd. (though I'd love to see one.)
> 
> The gnbd sources aren't part of the current gfs release, yet.
> 
> At the moment, they can be downloaded by the gfs's cvs repository by either 
> using cvs (follow instructions at 
> http://www.globalfilesystem.org/software/cvs.html ), or browsing the repository 
> at http://www.globalfilesystem.org/cgi-bin/cvsweb/GFS2/src/gnbd/ .
> 
> Besides, as long as I know, it is mostly like the linux nbd driver (so, I had 
> it stright on my nose...), but with few enhancements to gain a better 
> integration with gfs.
> 
> I believe it is still NOT the best solution to mirror disks by network: it uses 
> a tcp connection giving an emulated block device on a side, which is attached 
> to a real block device on the other.
> 
> A better fs-independent networked raid solution should be scalable, allowing 
> more than two hosts to cooperate in the creation of a 'virtual' raid. This
GNBD allows more than one client. 
But it suffers from consistency problems (at least this is what i figured
out in my 1 min look at the code).
> would be better implemented using (possibly multicast) udp messages. Here the 
> matter is complicated by the need to keep in sync every local image of the raid 
> (srv A writes and sends an update, srv B looses the update and wanna read), but 
> It should be made at the md or raidX layer in Linux, not below...
Nod. I will send the RAID1 write-only patch tomorrow which at least makes
the NBD+RAID1 more faster.
Jeffrey sent me his NBD kernel server code today. Tomorrow i'll take a
look at it and i already added using this to make a GNBD kenrel server in
my TODO list.

  - Marcelo