Remote Replication - Please help me understand how it works

[dartec]

Hi All,

I am new to TrueNAS.
I need some help understanding how remote replication works.

This is what I currently understand on how the process works:
1. A snapshot is created of the dataset/pool, etc. on the source
2. That snapshot is then transferred to the the destination machine

I also assume:
Even though the data is now sitting on the destination machine, it is not accessible on the network (as no sharing protocol has been configured for it)
If I lost the source machine and needed to restore the data (to a new machine) from the destination machine, this would be possible.

Is my understanding of the above correct?

TIA

 

[sretalla]

1. A snapshot is created of the dataset/pool, etc. on the source

Right.

2. That snapshot is then transferred to the the destination machine

To be clearer...

All the blocks referenced by that snapshot and its parents are transferred in addition to the snapshot itself.

Even though the data is now sitting on the destination machine, it is not accessible on the network (as no sharing protocol has been configured for it)

True. And depending on the settings you selected, possibly marked read-only.

If I lost the source machine and needed to restore the data (to a new machine) from the destination machine, this would be possible.

Yes, just a replication in the opposite direction to the new machine (and maybe marking as not read-only as part of that if necessary).

Is my understanding of the above correct?

Mostly. Feedback provided.

 

[dartec]

Thank you for this.

So if I want an exact copy of my data from the source machine, on the remote machine, would I set up a periodic snapshot task and then a replication task to transfer that snapshot to the remote machine?
When the new snapshot is replicated, would it just "update" the data on the remote side, so that I only have one copy of the data?

 

[sretalla]

So if I want an exact copy of my data from the source machine, on the remote machine, would I set up a periodic snapshot task and then a replication task to transfer that snapshot to the remote machine?

You can either setup a snapshot task or have the replication task handle the snapshot creation itself, up to you.

When the new snapshot is replicated, would it just "update" the data on the remote side, so that I only have one copy of the data?

Snapshots replicate changes to the source datasets incrementally (I think that's what you're asking). The same data is only copied once, with new data being copied as it is created.

You can select in the replication task how you handle deletions (matching the source or not) of snapshots.