philipp.reisner at linbit.com
Mon Dec 3 11:50:40 CET 2012
After 18 month of work we are happy that we finished the groundwork
for the next generation of DRBD.
In this development effort we moved all connection related attributes
from the resource object to a dedicated connection object.
A resource can have multiple connections from now on. The advantages
of multiple connections over stacking are:
* IO is tracked by one activity log (with stacking two ALs)
* scales to 31 peers (with stacking to two peers)
* Improved handling
The (optional) automatic promote feature opens new ways to integrate
DRBD with other cluster managers than pacemaker. It is also
useful to keep the dual-primary time short for live-migrations of
virtual machines in pacemaker clusters.
Last not least the improvements in the activity log code allow unseen
number of IOPS for DRBD replicated storage if the storage is larger
than the area covered by the AL.
Please keep in mind that this is the first pre-release, it will be
easy to find and trigger bugs. We are aware of that and will release
improved follow releases regularly. -- Please do not send bug reports for
this pre1 release.
* Support for multiple connections in a single resource; this feature
obsoletes device stacking; new configuration keywords: connection,
host, node-id, connection-mesh, hosts.
* Automatic promote; if a process opens a drbd device for r/w, drbd
tries to promotes that resource to primary; it the last process
closes the device drbd demotes that resource to secondary; opens
fails if promotion fails. This feature obsoletes become-primary-on
* Activity log striping; Besides the striping this allows AL
sizes of up to 65536.
* Non blocking queuing of AL-updates; This change significantly
improves the number of IOPs in case the workload does not fit into
the configured AL size.
* Resync extents are now 128MiByte instead of 16MiBytes
More information about the drbd-announce