I'm a R&D plebeian; not my job. We have a lifecycle management team that monitors metrics---e.g. system availability, failure rates, spares inventory, obs, etc.

I'm a R&D plebeian; not my job. We have a lifecycle management team that monitors metrics---e.g. system availability, failure rates, spares inventory, obs, etc.---and postures as risk/budget considerations permit. In the aggregate, these legacy systems are on the right tail of the bathtub curve quickly approaching EOL, hence their modern replacements already being fielded. As legacy systems are decommissioned, reclamation buffers the spares inventory pool. Push come to shove, cannibalization in the field keeps most balls rolling at the cost of diminished availability/throughput. Based on a sample size of 2 legacy systems in our lab that saw daily use over the past decade, things definitely broke, but I don't recall the production Alpha hosts every giving up the ghost beyond CMOS battery replacement nuisance. We still have a few support Alphas laying around just in case, but virtualization[1] on commodity bare metal effectively renders them redundant. [1] https://www.stromasys.com/solution/charon-axp-alpha-virtualization-and-emulation/ https://www.stromasys.com/solution/charon-axp-alpha-virtuali...