As a quick example, let’s go back to the olden times – back before we all had grey hair, coke-bottle glasses, and walked with a hunched back and cane – 2005-2010. Back then, it was very common to have a separate physical server for every workload. When that was the case, it was also common to have those servers purchased at different times to stagger their replacement costs.
For example, one might buy four servers in 2005, four in 2006, and four in 2007 with the plan to replace those first four in 2010, the next four in 2011, and the next four in 2012. Essentially, we could budget a similar cost every year and go through our budgeting machinations.
Then virtualization came around, and suddenly we needed significantly fewer servers. Maybe now we only have three physical servers, so we just used the newest three for VMware, the other new one for Veeam, and then bought a new SAN.
Wow, we just saved a lot of recurring money. However, we ran into a new problem – infrastructures don’t quit growing. When the time comes to replace those physical servers, we need to replace them with considerably larger physical servers that cost more money. So, maybe we stagger them and only replace one or two a year, and then replace the SAN for a really big expense every five years. We’ve still staggered our budget cycles a bit, and can continue budgeting roughly the same.