Post

On the eve of Kubecon I think it's time we acknowledge the cloud hasn't lived up to the hype It is an amazing tool that provides a ton of convenience, but long term the promises turn out to just be marketing hype

5 reposts50 likes

I've been skeptical of Kubernetes for a long time for the same reasons, but recently I started reconsidering. Self-managed k8s is less of a pain than it was five years ago, and - if well thought - it can enable interesting levels of standardization/simplification.

I agree. Making self managed k8s dead simple is why I joined @siderolabs.com I’ve tried a lot of k8s products and Talos + Omni is the easiest to maintain But I also think for a lot of companies VMs are a better fit for what they need and are usually less complex

Iiiiiiiiif you can get and keep talent. And success disasters can’t happen. And you can weather outages. And you rock at cybersecurity. And capital is cheap. And you can predict needs in advance. And no innovation from three multi-trillion dollar multinationals could possibly help your business.

I've been seeing people realizing this a lot lately. As someone who started working with AWS in 2008, I completely agree. It was a game changer for small companies back then, but it makes a lot less sense today as the main production platform for most use cases

Hot take!! 🔥 but a good one. Someone at the Startup fest today said “for new companies, using the cloud is table stakes” - thoughts? It’s nice how quickly you can move on ideas and buildings with the cloud

If you don't know what your revenue or needs are for at least a year you should use the cloud If you travel you should stay in a hotel. It's convenient even if it costs 20x per night as rent.

Out of interest, would you say this still stands up for smaller companies/startups OR companies without a team of people capable of operating servers. It's a skillset in itself operating infrastructure right? I agree with your points about complexity but I don't think salary of people is factored in

Just read the new stack article you linked, the point about the cost of a cloud centre of excellence is fair I hadn't fully considered. Id say the pool of people with that skill set is bigger though. Still interested to get your thoughts on companies looking to mo without infrastructure skills 🤔

The numbers in the article are general guidelines based on trends I’ve seen and companies I’ve worked at or with The thing I like about “ground computing” is “it depends” is a valid answer If you are small and can’t project your spend for a year or more then the cloud is a great option

But also, maybe a small team with 1-2 people who know how to manager servers is a good idea Look at Bluesky They’ve more than doubled the amount of users in about 4 months and activity breaks records every week They have a few hardware servers that handle the majority of the load.

It has totally lived up to the hype. It's one of the few tech advances that has in the last decade or so. Back to on prem is the new hype train, which for the vast majority of orgs, will not live up to the hype.

A significant reason for this being that, as with any technology hype, many will get caught up in all the perceived benefits without understanding the trade offs. Most engineers nowadays have no experience of running on prem. Will listen to DHH, copy big orgs, cargo cult.

I don't think it is absolute. The cloud has worked out quite well for many organizations. Not having to spend big chunks of my day managing on prem equipment has been nice. Of course it's not a panacea and doesn't fit for some teams. If you have very consistent workloads and enjoy tinkering..