Node Split

[pigmej]

In our ongoing effort to lower the bar for smeshing, we’re exploring a new direction for reducing the system requirements for smeshers. The idea is to re-architect the internal modules in go-spacemesh, isolating the smeshing logic from the passive consensus code. By dividing the node along those lines, we can enable users to run the lightweight, yet sensitive (access to private key) smeshing logic separately from the rest of the node. This can enable users with limited resources to use a remote node for their smeshing.

We want to introduce a node-split where we split node into multiple smaller entities as shown in the picture:

We believe that such a design will also be useful for larger smeshers, as it enables them to run small, lightweight smeshing services next to their PoST data. All the smeshing services that they run can then connect to a single passive node serving their entire operation.

That should answer the following problems:

• hard to run node at home with current network conditions, node requirements are bigger and bigger
• OpEx to run a Node is relatively high even in 1:n scenarios
• Should allow to hot-swap the nodes for better reliability

The overall expectation is that node still should be able to run "everything" like we today do with post-service in supervised mode.

DoD
There will be multiple phases of the activity.

1. very PoC where we prove that it will work Activation Service PoC go-spacemesh#6354
2. better PoC that will work more as anticipated
3. a proper solution.