You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Teranode announces Subtrees on a topic of the Libp2p private DHT. We should listen to that rather than full blocks from SV Nodes using wire protocol.
Motivation
We should be prepared for Ternodes taking over the network next year.
Description
Using Libp2p we should listen to the subtree announcements from Miners. Once we have the subtree URLs we should download the subtree list of txids to create an index from txid => MerklePath. When Metamorph registers a submitted transaction, we can then lookup the appropriate subtree to calculate its Merkle path. The discussion here can be around specific implementation which isn't crystal clear yet. We expect that it will be cheaper for a miner to facilitate this Merkle path service rather than an independent since the download would not be necessary.
Early discussions around this topic were around whether to pre-calculate the sub merkle tree paths or not since this would allow the work to be distributed across the ten minute period between blocks. This assumes everyone will request their merkle path on creation of a block / every tx will send a callback to the originator.
Another discussion was around the compound paths possible with BUMP where it was expected that Overlays might request a single callback for all 10,000 of their transactions per block, or similar. This would greatly reduce the bandwidth required to get people their paths, pushing the load out to application specific networks.
The text was updated successfully, but these errors were encountered:
Summary
Teranode announces Subtrees on a topic of the Libp2p private DHT. We should listen to that rather than full blocks from SV Nodes using wire protocol.
Motivation
We should be prepared for Ternodes taking over the network next year.
Description
Using Libp2p we should listen to the subtree announcements from Miners. Once we have the subtree URLs we should download the subtree list of txids to create an index from txid => MerklePath. When Metamorph registers a submitted transaction, we can then lookup the appropriate subtree to calculate its Merkle path. The discussion here can be around specific implementation which isn't crystal clear yet. We expect that it will be cheaper for a miner to facilitate this Merkle path service rather than an independent since the download would not be necessary.
Early discussions around this topic were around whether to pre-calculate the sub merkle tree paths or not since this would allow the work to be distributed across the ten minute period between blocks. This assumes everyone will request their merkle path on creation of a block / every tx will send a callback to the originator.
Another discussion was around the compound paths possible with BUMP where it was expected that Overlays might request a single callback for all 10,000 of their transactions per block, or similar. This would greatly reduce the bandwidth required to get people their paths, pushing the load out to application specific networks.
The text was updated successfully, but these errors were encountered: