Hello
I am managing a group of 20 xen domU which use the following storage
model:
On the filer server each server has a LVM logical volume like
/dev/vg/myserver.root-partition
Then the logical volume is exported with Aoe whith a commande
like
/usr/sbin/vbladed 1 2 eth1 /dev/vg/myserver.root-partition
On a separate server, the xen domU config files
contains a line like
disk = ['phy:etherd/e1.1,sda1,w']
Finally the domU itself contains the fstab entry to mount the root
partition.
This works well, but I find the settings over complicated. I was thinking
of migrating to a simpler model where the xen domU mounts his file system
over NFS directly from a directory in the filer.
So insteaf of having LVM --> Aoe --> Xen config file --> domU fstab, I
would just have to specify the mount point in the domU fstab like
filer:/myserver/root when I have to provision a new domU.
The domU themeselves are running php/mysql webapps used internally, and
do not get more than 10-20 concurrent users, so I don't they are very
much I/O bound.
Is there anything that would speak against such a schema, apart the topic
of NFS security ?
lg
Emmanuel
--
openForce Information Technology GesmbH
http://www.openforce.com/
It is a book about a Spanish guy called Manual. You should read it.
-- Dilbert
--
To UNSUBSCRIBE, email to debian-isp-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact
listmaster@lists.debian.org