storage: high availability/cluster

Hey guys,

I plan to get rid of some linux servers, I am still evaluating the possibilities.

Basically I want to use 2 machines in a cluster with manual failover. I had this setup with linux and drbd ( HAST equivalent, but can do more stuff) - hosting various containers and VMs. On hardware failure, I manually promoted the underlying devices to being master and then started the containers or VMs on the second machine. (manual failover is fast enough, we had weird problems with pacemaker - most important is that the dataset was in sync)

I have various questions in my head ...
What is the best way to achieve this in FreeBSD? HAST? - how is the performance? Glusterfs and ceph seem not very well supported, maybe moosefs? How about NFS - is there a clustered mode? iSCSI? Anyone with a production scenario? ggated? gmultipath? Also read about the BeAst setup - is this production ready? How can it be secured over an unsafe network without too much performance loss - ggated/iscsi/HAST don't provide encryption? (stunnel? openvpn? or a means of having a fast in-kernel vpn with pre-shared keys?)

Any ideas, suggestions, experiences are highly appreciated - also other ideas or possible solutions. Thanks
 
Back
Top