When Azure blew up it did so because it was cheaper, much cheaper, not so much because of the features it offered. They are nice, but for a lot of industries you sort of can’t use them because it’s a bureaucratic nightmare to upkeep the required exit plans if you do too much vendor lock-in. For some industries this isn’t an issue, but for many EU enterprise organisations it is. So anyway, the big cloud was cheaper, but since it’s been raising cost pretty steadily, and, because all those hardware houses lost customers and the survivors became cheaper, the scale has now tipped.
As far as operations go on the software side, I’m not sure the headcount is that different. We frankly tend to buy that from third party companies anyway, but the pricing difference isn’t too different between whatever you want to do. I’m personally a fan of doing it with your own staff, but it’s just such a challenge because your IT budget isn’t going to be big enough to do in a way that is resistant to people finding new jobs. The third party agencies don’t have this issue because they sell things like networking to a bunch of organisations, so it’s not just one or two guys for them like it would be if we did it ourselves.
But yeah. Big cloud operational costs are getting to where it makes absolutely no sense to use them, all costs included. When we draw up plans for hardware we have extra budget for when some net controller fails 5 years before it’s supposed to, that sort of detail. Otherwise you can’t make an informed decision. Judging by the local tech environment here in Denmark, we’re frankly even going to be late to the party of leaving big cloud because of costs. Likely because we weren’t using it too heavily in the first place.
I fully expect that the pendulum is eventually going up swing back. It’s not like we’re stopping our contracts on Office365 or Windows licenses anytime soon, and eventually we’re going to get offered the same sweet package deals that landed us in Azure and not in AWS to begin with. But for the next 5-10 years I think a lot of EU based non-tech orgs are going to leave big cloud over cost.
For basic server capacity, leasing or buying your own hardware is a competitor.
But for software higher in the stack, like collaboration or marketing platforms, there’s often no “on-prem” option. So a SaaS business just needs to stay close enough in price to competitors that the pain of staying is less than the pain of migrating.
If all your competitors are raising prices… well, so can you. (And vice versa.)
Azure price increases are ridiculous.
And Azure is not even the best cloud service provider compared with AWS and GCP. Azure documentation is horrendous and performance is subpar. Although I do think their storage is slightly cheaper than other the other two. But I think they might still win compared to GCP as they are pretty strong in enterprise sales. I keep hearing good things about Oracle cloud too but haven’t tried it yet.
As far as where we run things, well, no. Things aren’t better in azure than they were 20 years ago when it was run on a server in the basement. It’s not even cheaper per usage. But a lot of this comes to how you buy these things in enterprise. Our biggest expense is never going to be anything but the windows licensing, the rest is sort of just added on. Right now Azure is getting too expensive to compete, so we’re likely going to leave, but it’s not like it really matters for the organisation beyond the budget. We could run things on raspberry pies tied to a racoon who’d power them from the movement it makes while it attacks random people in the IT department and the organisation wouldn’t care as long as it worked and was cheaper than it’s competition.
The reason this is useful advice is that, as the GP notes, many think they will be locked in to selling below-market forever if they give their first customer a good deal.
This really does not apply to MS. They do not price anything below market rate, in some sense they set the market rate. (Or get to charge a monopoly/lock-in premium above it).
When we start seeing our license and support costs planning on almost doubling in the next 5 years, that have already skyrocketed in the past couple of years, it makes us start talking about shifting to a different solution.
The reason may not be what you expect. It's not that we're angry with Microsoft for raising their Azure prices. It's because they've raised them so high that the competition is getting to the point where the business case for spending the next 10 years on our own iron (renting space at facilities that sell that sort of thing) is turning green. The only way you can continue to increase prices 5-10% a year is if you have a monopoly on what you sell. Eventually it's going to cross into an area where not only the annual licensing but also the migration will be cheaper at your competition.