Ok, what are the alternatives? A fully kitted out server rack, switches, redundancy, CPUs, multiple drives, backup storage or tapes, backup batteries, own power supply, software and licenses can cost as much as a new mid sized car?
Asking for a friend…
At the scale of one of the top websites by daily active users, owning your own infrastructure is absolutely cheaper than just throwing it on AWS. At a more realistic smaller scale where you might exceed the bandwidth available for your own hardware, there’s also the option of a hybrid setup where your content is mainly hosted on hardware that you control and then it automatically scales out to AWS or similar when demand spikes.
There’s really tons of ways to make web apps and server infrastructure cheaper than just renting it from a cloud provider, but many orgs lack the vision and drive to do so and just fork money over to [insert hyperscaler here] and watch their app go down when that hyperscaler goes down. I really question this mentality especially when the same organization has constant discussions about not liking how large their cloud provider bills are
One could argue that reduced maintenance costs are a value from the cloud providers. E.g. when my AWS VM dies I can get a new one back in <10m (faster with automation). When my self-hosted server dies I need to have planned for that with a warm spare and someone needs to physically be connecting new hardware. AWS allows you to pay more but have a predictable constant cost.
But I think that you’re right that it’s lack of vision. Everyone’s following the VC-backed company path, where it doesn’t make sense to save money for next year because we’ll be selling some entirely new company then.
One could argue that reduced maintenance costs are a value from the cloud providers. E.g. when my AWS VM dies I can get a new one back in <10m (faster with automation). When my self-hosted server dies I need to have planned for that with a warm spare and someone needs to physically be connecting new hardware
Yes this is absolutely a value proposition that CTOs/CIOs/IT Dept Heads need to be considering. You’re not just paying for the VMs and storage, you’re also paying to outsource all of the hardware and some of the configuration work, however you still need admins to manage all of the VMs and configs. If that labor savings is actually enough to cover the immensely increased cost of cloud resources over local/colocated resources that you own (the infrastructure costs are pretty minimal in comparison) than awesome, more power to you.
I really think the biggest value is putting all of your baseline compute in hardware that you own, whether on-prem or colocated, them if you need bursts of resources place that in the cloud. With hardware you own you can spin up temporary VMs, you can keep legacy VMs around, you can fling data around with impunity. These are all tasks that have real costs in the cloud that they will happily bill you for.
But your owned hardware is a set quantity, so if you are rapidly hiring a thousand people or bringing in a new organization or have publicly facing services seeing immeense growth or anything like that and need more capacity immediately, you can’t. It can easily take weeks to bring more servers online even in a rush job, meanwhile the cloud can hand you capacity immediately. That’s the value of the cloud that’s being missed
AWS is incredibly expensive, if you’re hosting something like GitHub or Netflix on them instead of just owning the servers, you’re incredibly dumb
Ok, what are the alternatives? A fully kitted out server rack, switches, redundancy, CPUs, multiple drives, backup storage or tapes, backup batteries, own power supply, software and licenses can cost as much as a new mid sized car? Asking for a friend…
At the scale of one of the top websites by daily active users, owning your own infrastructure is absolutely cheaper than just throwing it on AWS. At a more realistic smaller scale where you might exceed the bandwidth available for your own hardware, there’s also the option of a hybrid setup where your content is mainly hosted on hardware that you control and then it automatically scales out to AWS or similar when demand spikes.
There’s really tons of ways to make web apps and server infrastructure cheaper than just renting it from a cloud provider, but many orgs lack the vision and drive to do so and just fork money over to [insert hyperscaler here] and watch their app go down when that hyperscaler goes down. I really question this mentality especially when the same organization has constant discussions about not liking how large their cloud provider bills are
One could argue that reduced maintenance costs are a value from the cloud providers. E.g. when my AWS VM dies I can get a new one back in <10m (faster with automation). When my self-hosted server dies I need to have planned for that with a warm spare and someone needs to physically be connecting new hardware. AWS allows you to pay more but have a predictable constant cost.
But I think that you’re right that it’s lack of vision. Everyone’s following the VC-backed company path, where it doesn’t make sense to save money for next year because we’ll be selling some entirely new company then.
Yes this is absolutely a value proposition that CTOs/CIOs/IT Dept Heads need to be considering. You’re not just paying for the VMs and storage, you’re also paying to outsource all of the hardware and some of the configuration work, however you still need admins to manage all of the VMs and configs. If that labor savings is actually enough to cover the immensely increased cost of cloud resources over local/colocated resources that you own (the infrastructure costs are pretty minimal in comparison) than awesome, more power to you.
I really think the biggest value is putting all of your baseline compute in hardware that you own, whether on-prem or colocated, them if you need bursts of resources place that in the cloud. With hardware you own you can spin up temporary VMs, you can keep legacy VMs around, you can fling data around with impunity. These are all tasks that have real costs in the cloud that they will happily bill you for.
But your owned hardware is a set quantity, so if you are rapidly hiring a thousand people or bringing in a new organization or have publicly facing services seeing immeense growth or anything like that and need more capacity immediately, you can’t. It can easily take weeks to bring more servers online even in a rush job, meanwhile the cloud can hand you capacity immediately. That’s the value of the cloud that’s being missed