Convert two dual-controller AFF200 into MetroCluster

[shadow stream]

Hi, I've read that its possible to convert two separate dual-controllers AFF200 into MetroCluster. We want to do this because of testing vMSC in lab before considering it into production:

https://core.vmware.com/resource/vmware-vsphere-metro-storage-cluster-vmsc

Currently the nodes operate independently in switchless cluster configuration. Data will be served to ESXi servers on FC. We have two Brocade FC 6510 SAN Switches. Do we need to buy some equipment or just follow procedures to build from scratch four node fabric attached MetroCluster:

https://docs.netapp.com/us-en/ontap-metrocluster/install-fc/index.html

Do we need some specific license? We have SMBR-C, SNAPCTR_STANDARD,
VE_2, CIFS_2, FCP_2, FLEXCLONE_2, NFS_2, SNAPMANAGER_SUITE_2B, SNAPMIRROR_2, SNAPRESTORE_2, SNAPVAULT_2, ISCSI_2

Thanks for help!

[wicked yarrow]

You would need to re-initialize both clusters. So not really possible without major disruption. Also you need ATTO bridges. And you would need 4x FC-switches (like Brocades).

[royal osprey]

you can't convert an HA system into MCC without wiping. And the 2nd problem is, A200 was never supported for MCC

[wicked yarrow]

Oh and AFF A200 is not possible with MCC-FC 😄

[shadow stream]

Yeah destroying it and building from scratch was in plan 😄 But I didn't read that A200 does not support it. So well no lab for me! Thanks guys!

[royal osprey]

technically it would be interessting if you can create an MCIP with shared Ports (CI and MCIP like A220) even when its not supported (maybe Model is blocked in Code during the "metrocluster configure procedure) - but only MCIP should work with the internal disks and Software iWARP. No chance to build an MCFC. For an unsupported Lab, i would give it a try 😄

[wicked yarrow]

Question is what do you want to learn. For production workload you most likely would get a MCC-IP and not MCC-FC. The layout between these are really different even though the experience in daily-business would be mostly the same for the admin and end-user.
I would suggest to try SM-BC. That can be easily build with your 2x A200 without any disruptions to your current workloads.

The only limitation: Your A200 can go up to 9.11.1 max. So would miss all the SM-BC enhancements in the latest versions but for a POC it's enough I think.

Regarding licenses: If possible install ONTAP One then you have everything.
If not the only license missing for SM-BC is SnapMirror Synchronous. You can get the master key license here: https://mysupport.netapp.com/site/systems/master-license-keys/snapmirror-synchronous

[shadow stream]

We want to test site failure with VMware cluster and NetApp storage. One vSphere cluster with 2 nodes in one site and 2 nodes in second site. Then run some test vms in site A and kill it then check if everything will be online on site B - if vms will start there after site A failure.

We want to check if it will start automatically etc, with limited admin actions.

[wicked yarrow]

That's easily possible with SM-BC (or should I start saying SM-AS...). But you would need VMFS datastores with either iSCSI or FC. SM-BC is not possible with NFS (at the moment).

[shadow stream]

We have FC now, we can do it in both sites with Brocade FC 6510

[wicked yarrow]

If only your storage in siteA dies, your VMs would not even crash and need to boot up again in siteB. The storage in siteB would simply take over and you would mostly likely not notice anything. But for this automatism you need a Mediator in siteC to eliminate the risk of split-brain. The storage in siteB needs to figure out if the storage in siteA really died or if it's simply a network issue.

Start here: https://docs.netapp.com/us-en/ontap/smbc/index.html

[shadow stream]

But in case of complete siteA failure will it be able to start vms on siteB? Ofcourse with vCenter HA and vSphere HA solutions. VMFS will be seen in siteB and put online by mediator? So we don't need to register VMs manually etc.

[wicked yarrow]

Yes, that would work.

As mentioned in the other thread, check the solution description from VMware: https://kb.vmware.com/s/article/83370

At the bottom there are test scenarios and how SM-BC would behave during a storage failure, site failure, etc.

[shadow stream]

Thank you OG1, you are true OG! 😄

[wicked yarrow]

Your welcome

[shadow stream]

One last thing - article mentions vMSC so MetroCluster. But I think that just replicated SVM and its LUNs containing VMFS with SM-BC will just work?

[gentle phoenix]

...and I guess you will also need some disk shelfs, because the internal disks in the shelf cannot be used in a FMC... thet need to be connected via the ATTO Bridges...

[wicked yarrow]

No, the term vMSC is only the general term by VMware for this solution. You can implement it with many different storage vendors and different technologies. vMSC (VMware term) does not automatically mean you would need to use MetroCluster (NetApp term).
Here are all the possible solutions, with NetApp it is MetroCluster and SM-BC: https://www.vmware.com/resources/compatibility/vcl/partnersupport.php

[shadow stream]

Thanks for clarification. Using similar terms for slightly different solution will make my hair even more gray!

[subtle ledge]

So here’s the thing - what would you like to lab? The tech is rock rock solid.