Re:The end-to-end refresh of our server hardware fleet
by hueving on Friday March 10, 2017 @10:30PM
(#13844055)
I misinterpreted what you meant by a "short lived experiment". I took that to basically mean any project when you're starting out. My apologies.
Whenever I'm experimenting I rarely need a burst of 100 instances, it's usually 1 or 2 instances to run things and I prefer to run them on my own hardware.
For companies that have instances running long term it can very well be cost effective to own the hardware. My email server, web server and DNS server are on my own hardware with a co-location facility that I trust.
But for experimental stuff where you need to spin up a hundred machines for an hour or two you just can't beat the cloud (and that's my only use case for the cloud, though I can see others go much further).
I don't like the monoculture any more than you do, but to see this as me having given the 'keys to the kingdown to Amazon' is several steps too far.