Question, what are the "Related services" ? witness? and RC plugins? Basically, everything that does not involve history and usual blockchain API requests? Or something else I am missing?
I am thinking on building a new mobo, with 128GB RAM at home and I am trying to understand if the seed/witness/others could be on a node with less than 128GB, say... 32 or up to 64GB, and then use the 128GB one for the history (HAF) and normal API stuff.
Then get a proper firewall, proper phisical links isolation, etc, and serve Oceania/ASIA islands zone, since from here access to Europe kind of "sucks" LOL... no one can cut off this crap distance latency, and until I get Elon Musk cheap Starlink satellites, it's still going to take a while.
Studying this properly before I regret doing something weird =) And grabbing enough experience to how the API is attacked to understand how to manage myself this.
Related services above doesn't refer to plugins but to other non-node related infrastructure type projects. I just realized I didn't specify what they were running on (they are not part of API or Seed).
The main node (API + Hivemind) doesn't need 128 GB and can be done with 48 GB. You may go with 128 GB but that would be overprovisioning. Definitely a good idea to launch out of Asia.
Thanks, that helps. Still need to run my own experimentation but I will be targeting for HAF and if by then hivemind is running with the haf, I will run it too.
48 GB is way better and I can sustain that easy. Do you know if with the HAF that might get lower or higher? ping @blocktrades? or anyone else that knows.
HAF should require less RAM than an existing full node, because there will be no need to run an account history node (this will be replaced by running hafah on your HAF server).
👍