[DRBD-user] SAN, drbd, ha, shared disks, ataoe, iscsi, gnbd

Carson Gaspar carson at taltos.org
Wed Aug 13 19:17:16 CEST 2008

Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.


> Alex wrote:
>> I read about lustre (from SUN Microsistems). It seems that is well
>> supported on linux (centos5/rhel5) and has support for raid/lvm/iscsi,
>> is scaling well and is easy to extend. Is that correct? Using lustre
>> can i join all volX (exported via iscsi) toghether in one bigger
>> volume (using raid/lvm) and have a SHARED STORAGE?

You really need to tell us more about your use case, but yes, for what I 
_think_ you're doing, you want Lustre (or something like it), not DRBD + 
iSCSI + LVM + GFS + a pony.

I suggest reading the whitepapers on Lustre's web site to see if it will 
work for you. But think of Lustre as NFS where the back end is striped 
across multiple, optionally redundant servers. The only reson I haven't 
deployed it yet is that we need Solaris client support, which is a long 
way off (Solaris _server_ support is coming very soon, if it isn't here 
already).

If you _really_ want to use the rickety tower of 
DRBD/LVM/iSCSI/GFS/etc., I'd suggest:

Server A: Pool A, Pool Z
Server B: Pool A, Pool B
Server C: Pool B, Pool C
...
Server Z: Pool Y, Pool Z

Where Pool A is a DRBD mirror between Server A and Server B (optionally 
with LVM underneath to concatenate disks)

The cluster should make IP address A and iSCSI A available on the same 
host where Pool A is Read/Write

Clients can use iSCSI targets A..Z as GFS storage. If any one server 
fails, the iSCSI device should survive as it's storage and IP address 
will be failed over by the cluster manager.

-- 
Carson



More information about the drbd-user mailing list