Hi all,
I would like to follow up on the topic where a group of node providers each operate a small number of nodes distributed across a common set of data centers.
Background
- The current Gen2 reward structure unintentionally incentivizes deploying few nodes across several data centers. For example, three node providers operating one node each in three different data centers receive higher rewards than if each provider operated three nodes in a single data center.
- The group of node providers “Maksym Ishchenko,” “Ivanov Oleksandr,” “Vladyslav Popov,” “Bohatyrov Volodymyr,” and “Artem Horodyskyi” have each deployed a maximum of one node per data center across a shared set of 13 data centers. In total, this group provides 35 nodes (7 per provider).
- For example, 5 of these nodes are part of the SNS subnet, and 4 are part of the NNS subnet.
- As confirmed by the node providers, these nodes are co-located within the same rack in each data center. This poses a potential security risk, as the node providers may have physical access to each other’s nodes across multiple data centers — although the data centers have additional safeguards in place to restrict access.
- Please note: The target topology already includes the node characteristic “Data center”, so having multiple node providers in a single data center is not an issue. The observed issue arises only when the same set of node providers share nodes across the same set of data centers, using the same racks in each.
Short-term Suggestion
- I propose grouping these five node providers into a single node provider cluster. The node-to-subnet tooling would then ensure that only one node from this cluster can appear in any given subnet.
- This change would only affect subnet assignment logic and would not impact rewards or introduce any other implications for the involved node providers.
- As a result, network security would improve without disadvantages for the participating node providers.
- We discussed this suggestion with the node providers in question and they are fine with it.
Mid-term Suggestion
- When reviewing the Gen2 reward scheme at the end of the current period, we should address this observed side effect of the reward scheme.
- In particular, we should consider adjusting the reward reduction-style calculations (as function of the node count) as this created the incentive for the current setup.