Around 60% of our virtual machines currently run on older B2ms or B2s sizes with temporary disks, and those sizes are no longer available for reservations. We mainly use these machines as lightly loaded web front ends, so we do not need especially high performance and are trying to keep costs under control. I have started moving them to B2s_v2 and similar sizes. These newer options do not support NVMe, which is fine for our workloads, but I am wondering about the longer-term implications. Could SCSI-based VMs eventually become difficult to maintain or require deleting and recreating the machines? Would another VM family be a better choice if we want to avoid another migration in a year or two?
2 Answers
For lightly used web front ends, moving to the v2 B-Series sizes should be perfectly reasonable. NVMe is mainly important for workloads that need very fast local temporary storage, so its absence should not matter here. There is no obvious reason to expect SCSI support to disappear soon, and choosing an NVMe-based size would not automatically make the deployment more future-proof. Unless your workload changes significantly, the v2 sizes are a sensible, cost-conscious upgrade.
SCSI is still a standard and widely supported attachment type in cloud infrastructure, so there is no practical reason to plan around its retirement. Providers generally keep supporting it when customers still depend on it. I would choose the VM size based on CPU, memory, disk, and network requirements rather than trying to predict a future NVMe-only transition.
That matches what these machines are doing. My main concern was whether SCSI might eventually be phased out, since our team is small and I would rather pay a little more now than repeat the migration later. The current change is only necessary because the older B-Series sizes can no longer be reserved.