Hello Danny, I am Henry and I want to share my insight about your concern.
My approach recommendation is a side-by-side migration, where you build a completely new, parallel environment and then migrate the data and roles to it. We will use the Storage Migration Service (SMS). It is a feature included in Windows Server.
You can refer the Migration Plan Using SMS:
- Build New Environment: Set up your two new servers with Windows Server 2022. Configure your iSCSI storage and build the new Failover Cluster.
- Deploy SMS Orchestrator: Install the Storage Migration Service role on a separate management server (or one of the new cluster nodes).
- Inventory: Use the SMS orchestrator to connect to your current Server 2019 cluster and inventory all the file shares, data, and configurations.
- Transfer Data (Pre-seeding): Initiate the data transfer. SMS will copy the 50TB of data from the old cluster to the new one. You can run it for days or weeks, and SMS will perform incremental syncs to keep the destination up-to-date.
- Cutover: During a planned maintenance window (e.g., a weekend), you will perform the cutover. Performs a final, rapid incremental sync.
- Takes the File Server role on your old cluster offline.
- Moves the network name and IP addresses from the old cluster role to the new one.
- Brings the shares online on your new cluster.
- At this point, your users are now seamlessly connected to the new hardware without any changes on their end.
Answering Your Specific Questions
- Should I install Server 2019 or Server 2022? You should install Windows Server 2022 on the new hardware.
- What file transfer performance should I expect? Your previous experience (35TB in 48 hours) is a useful baseline, but you should not assume it will be the same. The performance will be determined by the bottleneck in the new chain, which could be:
- The read speed of your old Fibre Channel storage.
- The bandwidth and latency of the network link between the datacenters.
- The write speed of your new iSCSI storage.
You can use the Storage Migration Service to create a test job for a single, large share (e.g., 1-2 TB). Let it run and measure the actual throughput.
I hope this information is helpful.