Figure HA-12 · Storage
Storage options for rConfig HA
From shared storage to DRBD to replication, and what each can lose
rConfig
Active node
Passive node, used after failover
Replication
01
Highly available NFS
Node A
ACTIVE
Node B
PASSIVE
NFS
controller failover
controller 1 · serving
controller 2 · standby
Both nodes mount the same export
RPO: none
02
GlusterFS / cloud file
Node A
ACTIVE
Node B
PASSIVE
Replicated volume
GlusterFS, EFS, Azure Files
brick 1
brick 2
brick 3
or zone a, b, c
Every write lands on every brick
RPO: none
03
DRBD
Node A
ACTIVE
Node B
PASSIVE
Disk
mounted
Disk
not mounted
synchronous block replication
Pacemaker promotes and mounts
RPO: none
04
lsyncd / rsync
Node A
ACTIVE
Node B
PASSIVE
Disk
mounted
Disk
copy
copied after write
RPO: seconds to minutes