UNSOLVED
Determining data change rate for replication
I am using Celerra Replicator v2 on a couple NS40s with some virtually provisioned iSCSI LUNs that are connected to the ESX servers. We are currently replicating locally over a gigabit connection but want to try and replicate over a WAN connection.
Is there an easy way to determine how much data is changed on a daily basis to determine how much bandwidth will be necessary for replication? Would I have to use a checkpoint to determine this? Any insight would be helpful as I haven't seen anything in the whitepapers that answer this.
Thanks
Aaron
Is there an easy way to determine how much data is changed on a daily basis to determine how much bandwidth will be necessary for replication? Would I have to use a checkpoint to determine this? Any insight would be helpful as I haven't seen anything in the whitepapers that answer this.
Thanks
Aaron
- Take a look at the checkpoints of the FS with
$ fs_ckpt fs_name -l
It will list the checkpoints of a FS. If you look at the columns "ckpt_usage_on savvol" and the "creation_date" in it will give you a percentage of the savvol usage.
$ nas_fs -size ckpt_name will give you the ckpt_usage_on_savvol in percent and MB
If you add up the ckpt_usage_on_savvol from a point in time in history till today, this should give you an impression of the changed data - How about using Celerra Monitor? Can you can look at the "I/O History (bytes)" for the filesystem to determine how much bandwidth the writes use. I don't know enough about the Celerra to know if the replication will use more or less bandwidth than the bandwidth that the writes use.
Can you do a network capture on the Celerra network ports to see how much bandwidth the replication TCP conversation is using? Or how about using NetFlow to gather the statistics on that TCP flow?
aaronj1
35 Posts
454
0
Posted March 23rd, 2009 20:00
Thanks