July 6th, 2024

From Cloud Chaos to FreeBSD Efficiency

A client shifted from expensive Kubernetes setups on AWS and GCP to cost-effective FreeBSD jails and VMs, improving control, cost savings, and performance. Real-world tests favored FreeBSD over cloud solutions, emphasizing efficient resource management.

Read original articleLink Icon
From Cloud Chaos to FreeBSD Efficiency

The article discusses a case where a client transitioned from a costly Kubernetes setup on AWS and GCP to a more efficient solution using FreeBSD jails and VMs. The client initially faced issues with over-engineering and high costs in the cloud environment. By implementing FreeBSD jails and VMs on physical servers, they achieved cost savings and better control over their data. Real-world testing revealed that FreeBSD outperformed cloud solutions in terms of stability and resource efficiency. The client also discovered and mitigated security vulnerabilities in WordPress sites that were causing CPU spikes on Docker containers but not on FreeBSD. Ultimately, the client decided to focus on FreeBSD for future deployments due to its cost-effectiveness and performance. The success story highlights the importance of choosing the right tools for specific tasks and optimizing resource usage in IT infrastructure management.

Link Icon 2 comments
By @dafelst - 3 months
I'm a fan of FreeBSD, but nothing really in this article really is super specific to switching to FreeBSD, aside from the customers lucking out that the crypto-miner that was injected into their system wasn't able to run on there.

This feels more like an "experienced dev/sysadmin shows inexperienced dev team how to better manage their environments" type article. That in itself is a good message and worthwhile, but the FreeBSD stuff is a bit besides the point.

By @ksec - 3 months
>Kubernetes......

>they found a sufficiently powerful machine, with 128GB of RAM, 2 NVMe drives of 1TB each, and two spinning disks of 2TB each for less than 100 euros per month.

We could get a 192, or even 288 Core CPU in a single socket, depending on your Single Thread Performance requirement. And we are only about 12 months away from 256 Core with 512 vCPU. 8TB of Memory and NVMe Drives so fast that makes DRAM memory caching nearly irrelevant for most Web workloads.

I just dont understand why aren't people pushing more complexity or CAPEX to hardware. And instead pushes all the complexity to software which has a recurring cost.