One of axioms we’re supposed to accept in the absence of any supporting data is that public cloud is cheaper than doing it yourself, always and forever.
This is a myth which is passed off like it’s a law akin to the conservation of energy.
It’s more accurate to say that one type of cloud, be it public or private, can be cheaper than the other in specific cases.
I can understand why start ups jump on the public cloud, were I one I would too.
I see the quickest way to burn precious VC cash is to write a check to a server/storage/networking vendor when instead I could rent all of that by the drop and hire a few more coders to actually build something which will start generating cash for me.
Then there’s the added upside that new services are made available providing extended functionality, so the longer I stay the more new infrastructure options I get.
But as this Gigaom article discussing the topic with people who been there and back again shows, there could come a time where it no longer makes sense for you to carry on where you started.
This also applies to workloads you might always have been running internally, it could hit a threshold where it makes economic sense to eject it out into the public cloud from now until eternity.
If you work in IT it is your job to always get the most out of every dollar spent regardless of where you’re going to spend it. That means it’s up to you to get into the weeds on the numbers. Not somebody else or the CFO, you.
Are there massive cost saving to be made in shared architectures operated at scale? Absolutely. But that goes for the public and the private cloud. And lets not forget that in the private cloud those savings are passed directly on to you and aren’t skimmed off the top as healthy provider margins.
Kids of the public cloud folks aren’t going to school without shoes on their feet. If you think you’re getting anything cheap from a public cloud you can be damn sure it’s cost them a hell of a lot less than they’re selling it to you for, because if it doesn’t they’ll be dead soon. (Nirvanix)
But it doesn’t matter where you’re running your workload, always check the meter and if the meter is running consistently on the high side you can probably do it cheaper somewhere else.
If there’s a universal law in any of this that’s probably it.