They should be equally efficient.. one is for file level syncing, while the other is for block level. Also this question is better suited for the forums on www.nexentastor.org

~Anil

Thanks. What is the difference between the forums? I am using the paid version, not sure if that matters. I posted to the other forum as well, because I am confused on how auto-tier determines the changes with snapshots.

Hi Anil. I tried a follow up on the Nexenta.org forums, but no reply yet. I did some futher tests on auto-tier and auto-sync. Here is what I am seeing.

I basically took the same dataset, but in auto-tier I decided to use the "exclude" option to remove a pretty large folder that I did not need. I tried the same in auto-sync, but it does not exclude the folder.

Auto-Tier (single disk - gzip on destination - used exclude) 6000 sec for initial rsync 1300 sec for each additional 4gb at destination

Auto-Tier (mirrored disks - no compression - used exclude) 1700 sec for initial rsync 700 sec for each additional rsync 11gb at destination

Auto-Sync (no mirror - no exclude) 7500 sec for initial sync 500 sec for each additional 50gb at destination

This really does not seem to add up. Why would auto-sync be so much faster with so much more data? The destination machine is a Large EC2 instance running OpenSolaris 2008.11 snv101brc2 X86. I looked at "prstat" and load is very low.

Another strange observation is that "zpool iostat" shows disk activity on and off. For instance, it will read/write for a few seconds, then nothing happens for about 10 - 15 seconds at the destination.

Auto-sync is one of the NexentaStor provided data replication mechanisms. It is a schedulable, fault-managed, fully configurable NexentaStor Data Replication service that can be used in a variety of backup, archiving, and DR scenarios. Nature Cleanse