• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

NAPP-IT and Replication Question

D

DigitalDaz

Weaksauce
Joined
Mar 3, 2013
Messages
79
I currently have a setup, ZFS but not using Napp-IT made up of the following.

2U chassis with 8 hotswap bays, i3 processor with 32GB ECC RAM, 8x1TB Seagate spinners. 2x256GB Intel SSDs as L2ARC.

The box is the storage for a small ESXi setup. The ESXi and the storage box are connected using QLE2462 fiber channel cards.

Obviously my box is a single point of failure, its been running great for a good year at least now.

What I am planning to do is duplicate the above exactly, I'm not particularly bothered about HA and I definitely want to stay with ZFS.

I figure if I put Napp-IT on both boxes and get the replication plugin going on both boxes I can split the load between both boxes and replicate to each other. That way, I hope, I should only lose half my setup should one box fail and I should be able to pull the replicated volumes up fairly quickly and get them back to life. That's the plan, anyway.

I will probably do the replication by bonding 3 of the gigabit ports, there's 4 on these boards. I would like to use 10 gig for the replication but have no more PCIe slots

My 8 spinners are in 4 striped mirrors.

What do you think?
 
Only on the initial replication, the network bandwith is important.
The following incremental ZFS replications are quite small as they transfer only modified data. So 1 GB is not a problem when you run the replications regularly.
 
Gea,

You think the logic of splitting the load between the two is OK in a sort of active/active and then just pulling any failures up again and sharing back out? I realise I will have to make sure the GUID is the same.

I'd probably be replicating as much as is possible. 60 seconds if thats an option.

I'm also thinking rather than VMWare/Other backup method, I have maybe 20 vms max, if I created separate luns, I could just use send receive to my current back up server so each VM had a zvol, would that work?
 
That would work but I would prefer using NFS over iSCSI because of its simplicity and the shared use option. In all cased when you want to use the replication, you need to stop further replications, disable readonly and import the VM (browse datastore and right click on .vmx file -> add to inventory).

You you can also import the replicated VM in offline state.
Only problem: Replications might be corrupt as their state is like a hard powered off server - can be a problem depending of the moment of snapshotting.
 
Thanks for the advice, I'll obviously test a lot before I depend on it.

I was hoping, the datastore would just appear if using iscsi/fiber because of multipathing once I made it not read only and made a view.

With regard to the snapshotting, surely its as reliable as vmwares own tool or proxmox LVM snapshots.
 
ESXi snapshots adds memory state (you can restore a running situation) but it adds a delay during creation and every snapshot reduce performance. Every solution has its pros and cons.
 
You must log in or register to reply here.
Back
Top