Bitcoin is all over the news, this time with big investors pouring money into it, a global pandemic that put the economy on hold, and governments printing money like crazy. Price raises, it is a bubble again.
Maybe it is responsible this time to at least try to learn about Bitcoin without rushing to get it like everybody else seems to be doing again. The only way to learn about it, benefit from it and clear the doubts from the healthy skepticism is to test the technology by yourself. This is my experience setting up a node.
The most well known educator is this space seem to be Andreas Antonopoulos, and he has kindly made his work available for everyone with an internet connection. Over at that popular code sharing platform is his book Mastering Bitcoin . It is a nice and rigorous presentation of Bitcoin, and I decided to start learning from that book.
Right away from the start of the book you must download the entire Bitcoin blockchain. I first wanted to download the test network, it was quite a fast process on the VPS I booked for the task. Yet, I quickly realized that the examples used on the book use the main network and I had to give it a try. Host during blockchain download has a collection of graphs about what was going on my system. Downloading the testnet happened from 4:30 pm to 6:05 pm 2020/12/11. During that time my system performed OK, yet as you can see from the CPU utilization graph, a substantial amount of CPU time was wasted waiting for IO. The system seems to have loaded all downloaded data into the Memory cache and it didn’t take enough space in RAM for the task of indexing the blockchain. I read later on the Arch Linux wiki that during the indexing phase I should increase in the settings the amount of available, because it looks like the bitcoin daemon seems to have its memory usage capped on the default settings. This fact becomes more evident later on while downloading the main network. RAM usage seems to have an upper bound.
The download mechanism seems stable, yet at 11am of the next day 2020/12/12 there is a sudden change of behavior. The CPU utilization on userspace triples and the download network bandwidth drops to half. On the download logs that time corresponds to blocks starting on 2020/03/29. That is already this year. I’m yet ignorant if there was a big change since then on, if there were just a lot more transactions in volume to deal with or if that is the working part of the blockchain. Being that date so close after the big market crash of this year, I do assume it was the movement of people liquidating their bitcoin for cash. Now that money is on recent addresses and there is a lot more of reconciliation work.
It was an interesting experience, and I value having a monitoring of my host in place, otherwise this little insight I gained would have never been possible. I can start with questions to explore as I go further on learning about bitcoin.
It seems for now that to keep a node running you need about 2GiB of RAM, I needed 368GiB of disk space for the main network and another 32GiB for the test network. I have read that you can have a node that doesn’t keep the full history, I will try that later on. The CPU is high during the download and indexing yet later on it is really low, however I still need to test how it performs when doing queries.
The cost of running the node is very relative, I find that a VPS is cheap enough these days, yet it is a new expense every month and that adds up. Thus it becomes very relevant to be a participant in the network. On the other hand, even if you don’t. The fees banks charge nowadays are the same, so running your “own bank” as many say cost you the same in fees, only that you now need to put some extra work to manage your node. I’ll see where does this experiment take me.
It was also a fun experience to learn about uPlot to draw the graphs. It is a very small library, with just about the minimum to plot good looking time series. It was simple to embed on this website and I’ll most certainly use it on the future.
As scientist I studied the physics of the very small quantum world. As a computer hacker I distill code. Software is eating the world, and less code means less errors, less problems. Millions of lines of legacy code demand attention and have to be understood and simplified for future reliable operation.