- Buy new hardware
- Update your firmware and check for new Virtualisation options
- Install ESXi 7 on same hardware with AllowLegacyCpu Option
- Just keep running ESXi 6.7 with latest patches
- Rent an offsite, shared host
The most obvious option, that you are going to propose to management, is to upgrade the server hardware. Finance will naturally argue that upgrades aren’t in the budget, and services are moving to the cloud, and it costs less to purchase extended 3rd party support than new hardware. My suggestion is to turn on Disk Quotas and let a few finance managers run out of disk space; your upgrades will be approved immediately.
This should have been done long ago but there was always something with higher priority. This isn’t a $0 solution as it still takes your time, but it doesn’t take a month worth of RFP’s and vendor selection and PO’s and why can every manager go on vacation at the same time but we have to have coverage?
AllowLegacyCpu is a mixed blessing. It allows an upgrade of older hardware to newer OS so we can get updates BUT the hardware is not supported or tested so any of those upgrades can make our ESXi server unbootable.
Ignoring updates is a recipe for disaster so mitigation’s need to be put in place, which could cost as much in work-hours as just replacing the hardware.
Renting a shared host also has its tradeoffs; unexpectedly high bills, poor performance for I/O intensive tasks, extra bandwidth requirements. But could be a great plan for fail-over/DR/auto-scale as you rarely use the resources so the bill is little.
A reasonable solution is likely to be some combination of the above; add 1 new server to your farm, upgrade the existing servers using AllowLegacyCpu. Run critical services only on the supported hardware. Budget for more hardware, unless you are investing hard in cloud. Mitigate the risk of hardware failing by renting a shared host.