My hobby project

So, the main idea behind this blog is to write about what I've done and what I'm about to with this hobby project of mine.

In a nutshell, I want to learn about Linux, servers, networking, infrastructure and all the best-practises there is to set up and manage my own servers and infrastructure.

The inspiration

In my undergraduate studies I wrote a research thesis on occurrence graphs under the guidance of Henning Ulfarsson from Reykjavik University. Our work could be categorized as experimental mathematics because of it's heavy reliance on computer analysis. To get a feeling for our new objects and mathematical structures we often wrote simple programs that looped over millions and billions and even trillions (that's a whopping \(10^{12}\)) of permutations to do some simple computations and then aggregate the results over the permutations to a sequence of numbers, something that we have a much better understanding of.

What I realized then, and have not stopped thinking about, is how extremely parallell these computations were. Today, all the rage is about programming on the graphics processing units (GPUs) for parallel tasks. It's all well and good when you have a shared memory model, but what if all your parallel tasks can work completely on their own on completely different computers and needed just a simple and short job description? In my and Henning's scenario the job description would only have been a few bytes, given the other computers would have the necessary libraries to do the calculations. I think Bitcoin mining pools works this way.

My inspiration is to learn enough to be able to expand my computer and it's computing capabilities to another machine. Eventually I'd be able to scale my computer horizontally to more computers, over the internet, all over the world.

I want to poke around to see what's possible and what's not.

So I need another computer

My first step must therefore be to set up my own server that I can experiment with. And in line with my mathematical background, I want to do it the right way.

I've been tinkering with Linux servers for the last 3-4 years now. I own several Raspberry Pi's and other single-board computers that I've used, among other things, as web servers. I've earned some experience at my work for WuXi NextCODE where I often have to debug Linux server errors.

But I didn't want to set up another handicapped low-performing Raspberry Pi. Instead I chose to relocate my Intel NUC with a 15W Intel Core i5-4250U processor, 128GB Crucial M550 mSATA SSD, 2TB spinning HDD (taken from a Seagate's Backup Plus Portable Drives unit) and 16GB DDR3L 1600MHz memory from Crucial.

I put my NUC under the TV and plugged it into the router. In the following blog posts I'll write about how I wanted the server to turn out and what I managed to do in my first attempt.