Historical data snapshot Feature

[LucasMLK]

xdag currently stores nearly 340 million blocks. According to the design of loading all of them into memory, it will seriously increase the operating cost of the mining pool. This will cause xdag to become more and more centralized. We need to design a snapshot method to reduce the operating cost of full nodes.

requirements:

1. Export balance data specification

• At present, there is a -l parameter in the c code.

• The out_balances() function can be called to output the block address and balance.

Current format:

address	amount
++/H4OicSdhuO0Sebhcd4t03DQdFy4+Y	1024.000000000

Snapshot export format:

address	publickey	amount
++/H4OicSdhuO0Sebhcd4t03DQdFy4+Y	036d27f617ce7b0cbdce0abebd1c7aafc147bd406276e6a08d64d7a7ed0ca68f0e	1024.000000000

Publickkey export will prepare for the future deletion of address block, It is recommended that the public key be in a compressed format

2. Full nodes export a snapshot of all address balances at the specified height

The balance changes in real time, you need to stop at the specified height before exporting the balance

3. Full nodes can import balance snapshots and continue to run

Imported according to the balance snapshot, does not rely on the original 340 million block storage, and does not affect the original user balance

[Sofarlemineur]

Hi, Very interesting. Do you think that this kind of node could be run as node on a simple PC or does it need also the actual hardware specification? Thanks for your work

[swordlet]

Snapshot can't work on a running node, for balance must be fixed in snapshot working process. So a snapshot node would be developed. And it is a off-line node. How to using blocks height to cut the storage into two parts: before snapshot part and after snapshot part? New mining pool will load snapshot and after snapshot part of storage.

[LucasMLK]

Snapshot can't work on a running node, for balance must be fixed in snapshot working process. So a snapshot node would be developed. And it is a off-line node. How to using blocks height to cut the storage into two parts: before snapshot part and after snapshot part? New mining pool will load snapshot and after snapshot part of storage.

Yes, if a snapshot node is added, the newly added node only needs to synchronize the latest snapshot and the data after the snapshot.

[LucasMLK]

Hi, Very interesting. Do you think that this kind of node could be run as node on a simple PC or does it need also the actual hardware specification? Thanks for your work

It depends on the size of the snapshot and the index space required for data query. I think this means that an simple PC becomes a necessary job for a full node.