Notes: MCFC to MCIP Transition

** Work in progress **

A place for some notes on performing a transition from 4-node FAS8060 MCFC (switchless) to 4-node AFF-A800 MCIP.

Official documentation:

• Choosing your transition procedure
• We see FAS8060 to AFF-A800 is supported from ONTAP 9.8
• Note: From hwu.netapp.com FAS8060 is supported to ONTAP 9.8. AFF-A800 is supported from 9.5.
• And the procedure we will follow:
• Transitioning nondisruptively from a MetroCluster FC to a MetroCluster IP configuration (ONTAP 9.8 and later)

Note: "Switchover and switchback—operate on an entire MC 8 node configuration; DR workflows operate at cluster scope, so all DR groups in a MC configuration are involved."

Major Steps

1. Rack new hardware (AFF & Cluster/MCIP switches)
2. Cable AFF-A800 to Cluster Switches, and complete all other cabling (front-end and back-end)
3. Configure MCIP Cluster Switches (transition configuration, as it will be the cluster switch for the FAS8060)
4. Load software on AFF-A800, re-initialize as required, and leave on Cluster Setup.
5. Site survey (if not involved in 1 to 4 already) to check:
1. Verify installation of AFF and MCIP Cluster Switches is all good.
2. We have suitable and SFPs cables to convert FAS8060 to switchless
6. Convert FAS8060 to switched cluster (two sites)
7. Add new AFF-A800 nodes to cluster (two sites)
8. Complete AFF-A800 network configurations pre-MetroCluster setup (nothing data SVM.)
9. Configure new AFF-A800 nodes in a new DR-Group (complete MetroCluster setup)
10. (Optional) Setup encryption (see below) as required (may need some action on aggregates)
11. (Optional) metrocluster configuration-settings calibration measure (see below)
12. Configure AFF-A800 data SVM specific configurations.
13. Vol move to AFF-A800.
14. Decommission FAS8060 DR Group.
15. Finalize transition.

Note: Need to ascertain the level of documentation the customer will require for internal change control processes.

Note: Switchover, healing, and switchback operations during nondisruptive transition - MCFC + MCIP - negotiated unified switchover is possible. AUSO not available during the transition process.

How to verify ISL status

Cisco switch infrastructure:
Check ports online, no errors, status, SFP health.
show interface brief
show interface counters
show interface port-channel X
show interface transceiver details

Broadcom (BES 53248) switch infrastructure:
Check ports online, transceivers, RCF, errors, VLAN config.
show port all
show fiber-ports optical-transceiver all
show running-config
show interface counters
show vlan port all

Configuring Key Management

For MetroCluster it's essential that the DR-Group is configured first and before the Key Manager is configured. If not, Key Manager information cannot be replicated between all the nodes of a DR-Group. You will need at least one unencrypted aggregate per site in order to create the DR-group.

Other Notes

When setting up MCIP, stress test is part of the setup. The DR-Group would not complete without getting at least 4.5Gbps bandwidth per fabric. If the ISL is shared, the stress test may affect other services using that ISL.

metrocluster configuration-settings calibration measure

Example metrocluster configuration-settings calibration:

cluster1::*> metrocluster configuration-settings calibration measure

Warning: This operation will attempt to fully saturate the network connection to the remote cluster for measuring bandwidth. This may disrupt performance and should not be run when MetroCluster is under heavy load. Do you want to continue? {y|n}: y

Measuring latency and bandwidth between node "cluster1-01" (192.168.1.1) and node "cluster2-01" (192.168.1.3) over port "e0f".

Measuring latency and bandwidth between node "cluster1-01" (192.168.2.1) and node "cluster2-01" (192.168.2.3) over port "e0g".

Measuring latency and bandwidth between node "cluster1-02" (192.168.1.2) and node "cluster2-02" (192.168.1.4) over port "e0f".

Measuring latency and bandwidth between node "cluster1-02" (192.168.2.2) and node "cluster2-02" (192.168.2.4) over port "e0g".

Measurements complete. Use the "metrocluster configuration-settings calibration show" command to display the results.

cluster1::*> metrocluster configuration-settings calibration show

                           Home Latency Bandwidth

ID Subnet      Node        Port (ms)    (Mb/s)

-- ----------- ----------- ---- ------- ------

1  192.168.1.0 cluster1-01 e0f  0.221   14649 

               cluster1-02 e0f  0.233   16851 

   192.168.2.0 cluster1-01 e0g  0.356   18581 

               cluster1-02 e0g  0.257   15759