Cardano Staking Pool Moving from AWS to Home setup

My staking pool has moved from AWS to Home setup, specifically after the update to Cardano Node release 1.29

What I had previously was a AWS setup with t3.large, which has 2 CPU and 8GB Ram. But since upgrade to 1.29, and now 1.30, it seems all the nodes are using more RAM than before, and often exceed the 8GB limit that I have. This lead to the cardano nodes not operating normally, or offline from time to time. I was able to fix it by simply increase the AWS machine from t3.large to t3.xlarge, which double the resourse to 4 CPU and 16 RAM.

The problem is, since I am just a small pool operator, and have very small number of ADA staking in the pool, I have zero blocks so far, and all expense is out of my pocket every month. This upgrade of AWS from t3.large to t3.xlarge will double my monthly expense from about $300 CAD to over $500 CAD a month. Previously I was doing it for the learning, contribution, and fun of being a Cardano pool operator, and I know I won’t have much delegator for a long while, but at $500 a month, I cannot afford it. So I made the decision of going from AWS to my home setup.

I bought 3 used Dell Optiplex Micro, put in 20GB RAM, and a 256GB SSD Drive, and setup my nodes very similar to how I did on AWS. Bought a new Firewall to protect my nodes, a Raspberry Pi as the Bastion host for single point of SSH access to the rest of the nodes, and now it’s running smoothly at home. I spent about $1300 CAD in all these hardware, and sometime to set it up.

Please check out my updated Architechture page for the setup details.

My advise for anyone trying to host your own staking pool but without much ADA to start with, hosting at Home is pretty much the only option. I didn’t expect the hardware resource requirement would went above 8GB this fast, or else I wouldn’t start on the AWS route previously, this is a lesson learned.

And for those that thinking about Raspberry Pi, the one with the most ram is only 8GB. You might be able to do some configuration changes, and optimize your node to run under 8GB RAM, but I would think future updates to Cardano will only increase resources it need. Since Raspberry Pi cannot increase the RAM, it will probably end up not feasible soon. I hope Raspberry Pi will release a 16GB version, but it’s probably not on their roadmap, as Raspberry Pi are designed for smaller task, not CPU or RAM intensive.