I’m currently running an Umbrel full node on a Raspberry Pi4 with 2GB of SSD. My main purpose in running a full node is to do on-chain analysis for BTC. I’ve been using Umbrel’s RPC and mempool apps to get API data from my node, with mixed results. For any larger API queries, such as addresses with thousands of txids, my API requests usually time out. The Umbrel’s settings screen shows my RAM usage never exceeds 2.5 / 4.0GB during these queries. I believe the time-outs are because the Pi’s CPU can’t process heavier API requests like this. However, it works fine for lighter requests, such as block or individual txid data. So, I think I need to use hardware with a more powerful CPU.
Do you have any suggests for more powerful hardware on which I could run my node instead of a Raspberry Pi4? Does Umbrel work with Windows 10, for example, if I wanted to use a laptop?
Any suggestions are much appreciated. Thanks.
P.S. I know I can buy this data online, but it’s much more fun and interesting to use the raw data from my own full node to do my own analysis.
Install a linux on a laptop works fine. (e.g. debuan / ubuntu).
I re-used a laptop with broken screen as server.
It has been running already for >3 years without significant issues… and you have a build in UPS
@Billiant4ever UPS do you mean the battery (Uninterruptible Power Supply)? It’s economically sustainable (from a power consumption perspective) compared wit a Pi4?
Yes, it has a build in battery (hence my Uninterruptible Power Supply remark), meaning you don’t need to invest in another UPS.
Wrt to the power consumption… dunno exactly. I assume my laptop consumes slightly more than a RPI4. I never actually measured it. But keep in mind, if you look it up online, that for a laptop, that is the max consumption with screen on. In case of using the laptop as sever it will be significantly less, as you use it with screen off.
Is it economical…that depends on the setup and what you have already available, where you are located etc. If you need to purchase one just for this purpose I would not think so. Keep in mind that you’re talking bout <25€ energy per year
Not very open source and “do it your self” kind of feeling, but an alternative to buying new hardware and trying to get a database performant is to stream the blocks you want to analyze into Google big query.
Not sure if it’s up to date, but they link an example of how they put the data there. They have 1TB free tier for the data processing queries themselves.
