Cloud Migration Is Back (If You Ignore the Actual Numbers) (indiadispatch.com) 40
An anonymous reader shares a report: The cloud migration narrative that powered tech valuations during the pandemic is attempting a comeback, but the underlying data suggests a more complex story.
UBS's new survey of IT services reveals a striking disconnect between industry expectations and customer reality. While executives proclaim "2025 will be far better than what we've seen in 2024," their enterprise clients report having migrated merely 15% of workloads to the cloud, with the remainder presenting increasingly complex challenges.
The numbers are particularly telling: Growth rates for major cloud providers AWS, Azure, and Google Cloud have declined from pandemic peaks of 40-50% to 10-20%. IT budgets for 2024, meanwhile, are projected to be "flattish to up very slightly, maybe a couple percent," marking a significant departure from the explosive growth of recent years.
UBS's new survey of IT services reveals a striking disconnect between industry expectations and customer reality. While executives proclaim "2025 will be far better than what we've seen in 2024," their enterprise clients report having migrated merely 15% of workloads to the cloud, with the remainder presenting increasingly complex challenges.
The numbers are particularly telling: Growth rates for major cloud providers AWS, Azure, and Google Cloud have declined from pandemic peaks of 40-50% to 10-20%. IT budgets for 2024, meanwhile, are projected to be "flattish to up very slightly, maybe a couple percent," marking a significant departure from the explosive growth of recent years.
I know why (Score:3)
Good. I make good money migrating them BACK (Score:4, Informative)
Re: (Score:2)
There are even companies that outsources their network management to external suppliers like Colt. (Would not recommend)
Re: (Score:2)
Well this explains why you're such a fucking dumbass. You still think it's 1995.
Cloud based hosting solutions save money on paying your staff by orders of magnitude. They are also more efficient. I know because I helped deploy fleets of video on demand servers when I worked at [major cable company] and we found that hosting on AWS provided a 15-20% increase in traffic capacity for 80% of the cost of self hosting.
People on /. are literally dumber for having read your comment.
Re: (Score:1)
I see you are profiteering of the scam.
Re: (Score:2, Interesting)
I am not surprised. There are not that many workloads and situations were cloud-hosting actually makes sense.
I've actually seen the reverse already (Score:4, Insightful)
de-migration. I knew this was going to happen. Someone in the government gets a recurring cloud bill and doesn't want to pay the cost anymore. They come up with estimates of the cost to go back to on-prem and do so.
My view is that it's hard to come up with a case where the cloud saves much money for anyone. Even if you got rid of some of your IT support, you'd be stuck getting more cloud developers to do the same things you had your hardware guys doing on-prem - troubleshooting and monitoring.
The government likes the idea of cloud because they hated all those independent data centers and wanted them all concentrated under some central bureaucrat's control. But that isn't enough to make it happen in a world where line-item funding of programs happens. If you don't control the $$$, you don't control where and what the computers are.
Re: I've actually seen the reverse already (Score:2)
The government likes the idea of cloud because they hated all those independent data centers and wanted them all concentrated under some central bureaucrat's control.
If that was their thinking, then it backfired. Bigly. I've seen a few "encloudified" operations. And their owners had effectively lost track [xkcd.com] of where their stuff was being hosted.
Re:I've actually seen the reverse already (Score:5, Informative)
My view is that it's hard to come up with a case where the cloud saves much money for anyone.
There are four areas where cloud hosting saves money, in my experience:
1. E-mail hosting. I've got one or two clients with ten-or-fewer employees who still on-prem, and with no real compliance requirements, something like Axigen or Mailcow are serviceable, but if you're of ANY size and scale, letting Google or Microsoft host your e-mail is a much better idea than keeping on-prem Exchange alive.
2. SaaS, for finger-pointing reasons. If you have an on-prem EMR software, it's either ridiculously customized for some highly specific use case, or you hate yourself. Letting another company charge you and handle both security and liability is a much better idea for most medical facilities, and it's just one example of where part of the blame is 'finger pointing'.
3. IaaS, correctly configured for elastic loads. Many years ago, when Alexa was a traffic stat counting website instead of a music playing kitchen timer, the 5-year graph on IRS.gov was amusing. Around April 1, traffic shot up to some ten times the usual traffic, then April 16, it dropped like a rock again. Does it make sense to have enough servers to handle the maximum peak load for three weeks a year? Or, would it make more sense to have it on something like AWS, where one pays for the needed resources for the three weeks they're needed, and then contract the resources again after the rush ends?
4. New businesses who don't know what they need yet. As a new startup with no customers yet, will a moderately-spec'd Poweredge R750, a Fortigate 100F, and 500Mbit sync on a /29 be enough? Or should one get a $250,000 rack, a /24, and another $250,000 rack in a different time zone 'just in case'? While a $100/month AWS bill can certainly balloon to the sort of numbers where on-prem starts being cheaper, it's nearly impossible to tell what is needed when the company is just starting.
Of course, 3 and 4 are how The Cloud came to its current position. Some startups were successful, and thus are coming to the realization that renting someone else's computer is only cheaper when you need 10% of that computer, rather than 100% of it.
Re: I've actually seen the reverse already (Score:3)
There are a ton of variations for #3 - its the same problem as #1, actually. If you need scale but the problem is completely outside your core business, it makes more sense to pay for the cloud service than maintain your own. In both cases though you are completely outsourcing the problem, so comparing the relative costs of on-prem vs cloud infra is effectively someone else's problem.
For most purposes the argument for cloud vs on-prem is an IaaS discussion - and ultimately the main argument is elasticity.
Th
Re: (Score:2)
As long as the internet is reliable that seems to make sense.
When the internet fails then you start to love your local servers.
Local servers aren't just for EMR. If you work in the process industry you definitely don't want network disruptions or latency to corrupt your production.
Cloud saves money same way eating out does (Score:3)
So EVERYONE moved to the cloud, typically a hyperscaler. We're moving back a lot of infrastructure off of it because:
1. The costs went up
2. The costs are projected to go up even more.
For some, absolutely, cloud is the way to go. If you're a startup, it makes sense. If you have dynamic needs, it makes sense. If you're a 5 pers
Re: (Score:2)
My view is that it's hard to come up with a case where the cloud saves much money for anyone. Even if you got rid of some of your IT support, you'd be stuck getting more cloud developers to do the same things you had your hardware guys doing on-prem - troubleshooting and monitoring.
Indeed. Running your stuff in the cloud is actually much harder and requires more skills. On-prem, you just firewall anything off you do not need or do not even install it. In the cloud, you need to look at everything and many things will be surprising.
I know several enterprises that have looked at the numbers and decided to stay on-prem or to just put some non-critical stuff in the cloud to trick "management" into thinking there may eventually be a migration. And now with _major_ observable problems with r
Our company plans on going back (Score:4, Informative)
The company I work for, moved all of our workstations to VMware cloud. They have now realized that it costs much more, have signed up for a single point of failure, poor GPU performance for those who need it... The dream of shared hardware resources saving money is not really working as promised. They are now planning a move back to under the desk computers. I guess for the current state of HW technology, cloud has touched it's economic ceiling.
You must have gotten a good deal (Score:2)
The company I work for, moved all of our workstations to VMware cloud. They have now realized that it costs much more, have signed up for a single point of failure, poor GPU performance for those who need it... The dream of shared hardware resources saving money is not really working as promised. They are now planning a move back to under the desk computers. I guess for the current state of HW technology, cloud has touched it's economic ceiling.
Last I checked, the VMWare license cost as much, if not more, than good hardware. Hard to justify all that, especially if you've ever used VMWare remotely and saw how glitchy it is.
Re: (Score:2)
Check again. Broadcom has realized they screwed up massively and is trying to make adjustments.
Re: (Score:2)
Wait, what? They got rid of PCs (desktop/laptop) from people who need GPus and replaced them with Chromebooks that go to the web to get graphics done?
Re: (Score:2)
Three letters to explain that: MBA.
Re: (Score:2)
Three letters to explain that: MBA.
May be true in general, but not true in our particular case. The company is owned privately and run by the founders, who are very competent technically and not MBA material at all.
Re: (Score:2)
Interesting. Well, apparently they realized afterwards that they made a pretty bad mistake. That already puts them into the group of the competent, even if only at the low end.
Re: (Score:2)
Yes, they replaced our workstations with VDI machines that are in the cloud. We already had windows laptops to remotely connect to our under-the-desk workstations, now we connect to cloud-based machines instead. The GPUs, for those who need them, are in the cloud too, attached to the VDI. So now there is a dedicated pool of VDI machines that have GPUs and people have to connect to them if they need one.
Re: (Score:2)
Yeah, that makes sense, because migrating hardware PCs to VMWare makes no sense, especially if graphical capability is required.
The kind of cloud migration that makes sense, is headless, such as moving from on-prem database servers, to cloud-base managed servers. Or moving on-prem web servers, to cloud-based managed web hosts. In those cases, hosting on-prem requires you to manage a whole lot of equipment that is not dedicated to the task, i.e., database and web hosting. And in the cloud, you can scale up a
The cloud is a trap (Score:2)
Run away
If you ignore the numbers, yeah (Score:4, Insightful)
"Growth rates for major cloud providers AWS, Azure, and Google Cloud have declined from pandemic peaks of 40-50% to 10-20%"
That's not a reversal, that's a saturated market. Until that number is negative, it's not a reversal.
(I do expect it will go negative within the next few years, since could it, in fact, more expensive and less reliable. The only companies who will continue going to the cloud are those who have a mission critical app they can't operate without from a vendor who only offers continued upgrade and security patches on a cloud migration - at several times the cost. Why, yes, I do know such a company, why do you ask?)
Re: (Score:3)
Re: (Score:2)
Exactly! Also, the world we're living in, where a 10-20% yearly growth is reason for worries because it's not again the same 50% of a completely outlier period.
Re: (Score:2)
Good point. 20% grown could easily be a 100% price increase on a 40% loss of customers.
And sinking? (Score:2)
I expect a lot more loss from cloud. And speaking as a computer professional (including a B.Sc) and a long career on mainframes, PCs, and servers, let me repeat what I've been saying for years: there's no difference between "the cloud" and timesharing on a mainframe. And management is finally realizing this, and wants more control over what's on their systems.
Re: (Score:2)
Ah, but there's the allure that literally everything can be done from the CEOs laptop. Going back to mainframes meaning having more smelly nerds showing up at the office!
Re: (Score:2)
there's no difference between "the cloud" and timesharing on a mainframe.
On a fundamental level, there really is not. But this move from local computing to central computing, and the other way round, has now happened several times. Always driven by irrational hypes and marketing lies. I guess it will need to happen a few times more for people to learn. If they eventually do.
Re: (Score:2)
Re: (Score:2)
Probably. On the other hand, some cloud providers manage to screw up exceptionally badly all by themselves: https://www.cisa.gov/sites/def... [cisa.gov]
It keeps coming back and forth (Score:2)
On your marks, get set, blow. (Score:2)
Before you migrate to the Cloud.. Get Tor and ... (Score:2)
2025 will be far better... (Score:2)
Yeah? Don't count on it.
Yes, it is The Future according to management (Score:2)
Yes, it is The Future according to management....
My company's goal is to have 99% of our apps in the cloud within a year, with only the sweet, sweet sensitive data (PII, PHI, HIPAA stuff, etc) residing in a comparatively small number of tightly managed on-prem data centers. There's a super-duper-ultra-mega-hyper-pumpkin-spice secure pipe between so the cloud thingys can talk to the data thingys.
Some look easy to move, some look hard, but they fully intend to have 99% of them up there running on multiple, r
OpEx and CapEx perhaps? (Score:2)
A few years ago, I worked for a company that did a complete forklift to the cloud. 1:1 for each VM, bare metal host, you name it. Their AWS bill was just mind-boggling. However, the CFO was happy with it, because the moved away from CapEx to OpEx, and could show their investors that they were a lean, mean company that had no investments in depreciating assets, but used the Cloud where either one used the Cloud at their job, or they didn't have a job. Apparently this was a good thing since at company is
Cloud not always makes sense (Score:2)
Cloud makes sense for elastic load, or experimental workloads where capex is too risky.
But for established, well understood workloads? It's usually way cheaper to keep your own servers and rent space in a datacenter.