Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Fri, Jan 16, 2009 at 03:52:59PM +0100, Tomasz Chmielewski wrote: > Of course there is no such thing as protocol D in DRBD. > > Currently, local writes are only as fast as data reaches the remote side > (plus some TCP buffer tuning with protocol A, which won't make a big > difference in a long term). > > This can be a bit of pain if you: > - have a slow link, > - want to use DRBD for disaster recovery mainly, not for (fast) failover, > - have lots of gigabytes to sync, > - want the writes to happen fast. we have the drbd proxy, which buffers and compresses ... contact sales at linbit ;) > I was wondering if a non blocking, async protocol was ever considered > in DRBD? considered. yes. > So that local writes would be almost as fast as a local media allows; > anything which was not transferred to the remote side would be marked as > "dirty" and transferred when the time/bandwidth/conditions permit. yes. will be implemented in drbd 9. will NOT be available within the next few month, though. meanwhile, consider disconnecting your DR site, and connect only during off hours. you will get a re-sync of the changes during that time. also have a look at the before-resync-target after-resync-target hooks in drbd.conf, which would be used for example to snapshot a lower level device before it becomes synctarget, so you still have the consistent snapshot, in case the primary site is bombed out during the sync, as the sync target is inconsistent during that time. does that make sense? -- : Lars Ellenberg : LINBIT | Your Way to High Availability : DRBD/HA support and consulting http://www.linbit.com DRBD® and LINBIT® are registered trademarks of LINBIT, Austria. __ please don't Cc me, but send to list -- I'm subscribed