The cost of cloud

I have a rule that is simple, effective but also quite rude: if you can't deliver and maintain a 500 instances large infrastructure, same uptime and all, at half the cost of AWS by yourself (1 person) in 3 months using only open source solutions basically you should not have an opinion about this. You are just rationalizing your incompetence on this particular subject. Sorry to be this blunt but I am simply tired of listenting to people who can't do it explain that what I do every day can't or should not be done.

Hiring your own security guard is cheaper than paying an outsourcing firm.<p>Hiring your own janitor is cheaper than paying an outsourcing firm.<p>Building your own office is cheaper than renting one.<p>Doing your taxes with pen and paper is cheaper than paying turbo tax.<p>Making your own food is cheaper than eating out. Hiring a cook directly is cheaper than hiring sudexo.<p>I could keep going. But sometimes it’s not just cost. The biggest two values you get with AWS is 1) reducing time spent outside your business’s core competencies and 2) a vast ecosystem - 3rd party offerings, readily available devs, consulting services, and compliance services.<p>I’d add that for those having compliance needs. It’s not always as simple as rack and stacking infra. You have to use services that meet the compliance auditors needs.

If you run your own datacenter, there is also the opportunity cost of slowing down R&amp;D and new development work.<p>Let&#x27;s say the year is 2012 and Redshift is introduced, completely changing how organizations can generate insights from their data. Running your own datacenter? good luck waiting for the ops team to install something similar! It might take you a couple of quarters assuming they are already competent at it. On the Cloud? Press a few buttons and you&#x27;re off to the races.<p>velocity is a competitive advantage

I think the mistake is often made by comparing primitives. E.g. running my own RAID vs S3. Colo traffic vs AWS traffic.<p>But what about comparing the whole ecosystem?<p>Can you provide a self hosted granular access permission to your RAID? How hard is it to configure and maintain?<p>Will your colo deflect a DDOS attack?<p>When you run your own services, you have to reinvent so much it doesn’t seem to be worth it.

These types of posts rarely measure operational cost. Anybody can buy infrastructure and stick it behind an API. But can you make it fault tolerant with high availability and low latency. Can you do all of that and _still_ beat AWS&#x27;s costs? For the vast majority of customers the answer is no.<p>I used to work at a medium sized company, and they saved millions by moving to the cloud, and gained much better availability&#x2F;performance. It wasn&#x27;t even close, because that medium-sized company didn&#x27;t have the expertise to operate the service efficiently. They just bought off the shelf stuff from VMWare etc.. Plus, DR meant paying double.<p>disclaimer, I now work for AWS.

&quot;My rough estimate is that the unit cost of provisioning a service on AWS is about 3 times that of a competent IT organization providing a similar service in house.&quot;<p>The word competent here is doing a lot of weight lifting.<p>I know companies in which you have to wait _months_ for a small server to be allocated to your team. AWS does it in seconds

I don&#x27;t know why there are so many articles in this vain:<p>- use the cloud until you&#x27;re making so much money that you can afford to hire the sufficient talent to replicate cheaper.<p>fin

I&#x27;m sure people will disagree, but I&#x27;ve always viewed the main benefit of cloud is a company with no talent or time to do networking &#x2F; infrastructure early on, so they use the cloud, which requires less expertise. Once they grow and can afford network &#x2F; infrastructure talent, the switching costs are hard to justify, even with the 3x markup (according to the article).<p>I can see why it&#x27;s compelling, but for me running my stuff on other people&#x27;s servers in this day and age is concerning. Like many computing things, it really depends on the situation.

A crucial point is to ensure that you give yourself the ability to move from your cloud providers to on prem, and &#x2F; or to a different cloud.<p>A bad example: If you are with provide X and they change pricing structure and it impacts you 3 times the cost, you can move it without too much hassle.<p>Once you embrace all the proprietary and fancy features that your cloud provider has, then you are stuck and moving will be a nightmare.<p>compute&#x2F;s3 are easy to move. k8s should be easy to move but I have not tried it myself. Database hosting is easy to change as long as you are using an independent product, not a custom database your cloud provider has.<p>It is much more difficult when you have AWS&#x2F;Azure pipelines, AWS&#x2F;Azure Geo location&#x2F;manipulation AWS&#x2F;Azure proprietary scaling etc etc<p>Terraform is supposed to help here but in my experience with TF consultants it is not at all straightforward not compatible with a lot of AWS&#x2F;Azure offerings.<p>Then you are stuck and moving will be difficult and expensive. Which of course is the business plan for the cloud providers.

For those that didn&#x27;t read it, I recommend the thread on bandwidth costs from 5 month ago this article links to.<p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=27930151" rel="nofollow">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=27930151</a>

I managed a deep learning team at Capital One during the period when they dropped their own data centers and moved completely to AWS. I am 100% sure that they made the right decision. My current company also uses AWS. With modern corporate infrastructure with micro services, and buying vs. building many 3rd party services, it can make sense to go all in on Cloud.<p>Personally, I miss the days of monolith web applications that were relatively easy to host on a leased server. I continue to be a big fan of Hetzner and their hosted servers as well as their VPSs are very reasonably priced. Another thing that I like about Hetzner, OVH, etc. is that their bandwidth costs are also very reasonable so moving databases and monolith web applications to a similar service does not have to be a big deal.<p>I think that each company’s needs have to be assessed separately.

These 3x calculations for cloud almost always ignore a real TCO calculation for a real organization and real app deployment and rarely compare capability, flexibility and recovery which have real value, but have tough calculations.

Mid sized cloud companies like OVH, Scaleway, Hetzner or Infomaniak are the way to go for me.<p>But here&#x27;s is the point that&#x27;s completely missing from the article. Many choices in organisations are leaded by two wrong drivers :<p>- Career risk awareness : like we said in the good old days, nobody was fired for buying Sun Microsystems. Same thing applies for Azure and AWS. You&#x27;re not paying the bill so why looking for a cheaper alternative that might cost you personally a lot. - CV driven decision : on your CV it&#x27;s better and more valuable to have 3 years of Azure&#x2F;AWS than having 3 years on OVH&#x2F;Hetzner&#x2F;etc.<p>Finally, as a leader in an organisation, it&#x27;s always easier to follow the trend rather than trying to convince your coworkers to follow you to a more &quot;exotic&quot; solution.

Maybe tangent to this but I am really enjoying Cloudflare Workers w&#x2F;KV and Durable Objects. Currently running over 80% of my company’s infra on it and it’s a monthly cost under $50.

I wonder how the costs compare if you choose to go outside of the big three. Hetzner, OVH, Scaleway etc are so much cheaper and the network exit costs are much lower than with AWS GCP or Azure.

&quot;...3 times that of a competent IT organization providing a similar service in house.&quot; A competent IT organization is a real stretch for most organizations.

I figured a large portion of the blame could be laid at the feet of leadership; these people want to lead people, not run IT. Moving to the cloud gets rid of all those pesky power and environmental and raised floor and employing all those weird people…and if it’s more money, that makes my budget look good, too. If you save money, you’ll get less next year.

<i>&quot;I use AWS for a lot of things, but I strongly regard the cloud as just another tool, to be used as occasion demands, rather than because the high priests say you should.&quot;</i><p>He&#x27;s right. Evaluate your needs and use it if it makes sense. Not a very controversial opinion, I think!<p>The problem is, how do you evaluate your needs if you aren&#x27;t an expert in <i>either</i> self-hosting or using managed services (AWS)? I think you should treat it like going to a doctor: get two opinions from two different senior professionals in two different specializations. And definitely make an assessment based on real numbers. Try to get ballpark figures from similar-sized businesses about their costs (capex&#x2F;opex, infrastructure, staff) and requirements (expertise, time-to-market, FRs&#x2F;NFRs, regulations, etc). Building a business is a huge thing, and how you use technology can either be a hindrance or an accelerator, but it has to fit your use case.

I think there is room for hybrid cloud built with open source software.<p>Datbases and virtual machines tend to be expensive in cloud providers. Steady state workloads.<p>Object storge tend to be fairly cheap.<p>There is also cost of vendor lock in if you use propritary cloud technologies such as databases.<p>You can build open source infrastructure with hybrid solution. Probably less risk of downtime.

Have not seen this being mentioned yet, here goes:<p>Choosing a large cloud provider is often much greener than doing your own or using a smaller provider.<p>Carbon is a large cost.<p>GCP has been carbon neutral for over a decade, AWS and Azure have made big public commitments.<p>Sourcing clean energy for 1000s of servers is not always easy.

Just curious: if you are an enterprise, what do you think (besides cost) is a compelling argument in 2022 for not going to a public cloud? This is probably a Ask HN, so please do point me to it if you know an existing one.

I wouldn&#x27;t underestimate the cost of internal politics...

I dont think many are actually against &quot;Cloud&quot; per se. But Specifically AWS, Azure and GCP. Especially when you are only looking at like EC2 and wondering why they are 3x more expensive to other managed server &#x2F; cloud vendor even on Reserve instances with zero bandwidth bundled.<p>If you look at prices from Oracle Cloud ( ignoring whatever feeling you have against it ) than all of a sudden it is extremely attractive.

&gt; In particular, traditional legacy IT infrastructures are ridiculously overpriced. (If you&#x27;re using commercial virtualization platforms and&#x2F;or SAN storage, then you&#x27;re overpaying by as much as a factor of 10, and getting an inferior product into the bargain<p>Could someone elaborate? Is this saying something free like OpenStack would be better than something paid like VMware? Is that really common knowledge?

I agree that 99% of companies should not be buying their own hardware and setting up a datacenter in their basements. But: not all clouds are created equal.<p>The &quot;cloud&quot; can just be a blank Debian box on Digitalocean where you have root access _or_ it could be some obscure managed AWS service where all the technical days are abstracted away behind a REST API.

I really like how these claim in the post not backed up by anything.

a lot of time its not about saving money. you can&#x27;t even hire&#x2F;afford people if you are a small&#x2F;mid size company.