2014-06-11 02:57 AM
I am quite new to NetApp Filers having previously working on EMC/DELL/3PAR, so the concept of MetaClusters is a new one. I have recently inherited 4 x FAS Filers in two MetaClusters - 2 in a Stretched MetaCluster and 2 in a Fabric MetaCluster. All are located in the same site, and I am looking to change this setup to a single fabric metacluster utilising 2 x HA controllers on each site. I was wondering if someone could point me to a document on the implications of doing this? Will all data be lost on all SAN's? Is it possible?
What is the maximum distance for a Fibre Metrocluster? Am I better off converting the metrocluster type to a fibre metrocluster if this is possible on the existing systems?
I have found numerous documents on creating and best practices for MetroCluster, but nothing on destroying and recreating.
Any help is greatly appreciated
Solved! SEE THE SOLUTION
2014-06-11 06:10 AM
Shared switches are supported for selected configurations. Assuming yours is qualified, the actual task is pretty boring. There is no data loss and existing FMC can remain online.
I do not see any advantage (from NetApp side) in doing it. Distance limit is several hundred kilometers to my best knowledge, but there are other considerations as well.
I would not do it if you are new to NetApp - it is not rocket science, but having some experience helps. If possible, try to involve professional services from NetApp or partner.
2014-06-11 06:35 AM
I am looking to change this setup to a single fabric metacluster utilising 2 x HA controllers on each site
Not doable - you can have two MetroCluster setups sharing (potentially) back-end FC switches, but it will be still two HA pairs *spanning* sites. MetroCluster-type setup linking local HA pair to a remote HA pair is hinted in Clustered ONTAP 8.3.
What is the maximum distance for a Fibre Metrocluster?
Up to 200km depending on speed, cabling, and the number of connections.
2014-06-11 07:22 AM
Just to make sure I have it right,
I would keep the current NetApp metro clusters as is, simply migrate the stretched to fabric, making sure that I have qualified FC Switches between sites. I would then simply ship the second units to the DR site and reconnect and synch once they are up? Is that correct?
So in essence, I can have two seperate Fabric Metro Clusters replicating from Primary to DR using the same infrastructure (10GB, 40KM distance sub 1ms ping).
I will get a pro to come and do the work, I just want to get an idea of the design myself.