The Quest for Lightning Block Propagation

Here we see a Turtle having a bright idea.

TurtleCoin has 30 second block time which makes transactions on the network faster than other projects; however, it poses some challenges in making sure that new blocks propagate across the entire network as quickly as possible to prevent the creation of unnecessary network forks. While increasing the number of nodes on the network helps to achieve this goal, reducing the size of the blocks that are transmitted between nodes allows us to propagate a block so fast it’s almost as if they were transmitted by lightning.

Network Propagation

Block Propagation at its finest

Understanding what information is propagated throughout the network is a requirement for making sense of what we’re trying to accomplish when we talk about speeding up propagation.

Each node that is connected to the network performs vital functions such as, but not limited to:

  • Verifies transactions (signatures, no double spending, structures, etc.)
  • Verifies blocks (makes sure that they meet the difficulty requirements, are valid in the chain, contain valid transactions, etc.)
  • Maintains a transaction pool (transactions that have been submitted to the network but not included in a block)
  • Relays transactions & blocks to peer nodes (those the node is connected to)

Each time that a new transaction, in its entirety, is submitted to a node for processing by a user, that transaction is added to the node’s transaction pool and then relayed to all its connected peers who relay it to their peers until everyone on the network has those transactions.

Mining a Block

We know you crave finding those blocks

Once a transaction is in the transaction pool, it’s hash is then included in the block template that is passed to miners via pools or directly from the node if solo mining.

When a miner (or pool) finds a block the block, with all its glory, it is submitted to the node, it is validated, added to the chain, and then the full block containing every transaction including its inputs, outputs, signatures, etc. is relayed to all the node’s peers. Those peers relay it to their peers after verifying it and adding it to their copy of the blockchain, and so on until everyone has the new block.

Block Propagation Payloads

The full blocks that are sent between nodes have the sizes as reported on the block explorers. Some blocks are small at only a few hundred bytes while others with many transactions or transactions with lots of inputs/outputs can climb to well over 100KB. 100KB does not seem like a lot of data, but when you need 100KB to propagate between >1,000 nodes, it takes considerable time. We need the elapsed time from block creation to full network propagation to be as little as possible to make sure that the nodes that are mined against always have the latest information about the chain.

But… Why?

You may be asking yourself, “If transactions are propagated around the network for the transaction pool and a node already has them, why are we sending them again, in their entirety, with every block that is propagated?”

Things don’t have to make sense to work, right?

To put it bluntly, we don’t need to send everything all the time if a node already has most of the information needed to verify a new block. Sending all of it again is just duplicating things it already has.

Lite Blocks to the Rescue

Block Propagation is like a game of cooperative Tetris

Enter the concept of Lite Blocks. Lite Blocks are a special kind of block created just for synchronization between nodes. A Lite Block is just like a full block except it only contains the transaction hashes not the full contents of the transactions (inputs, outputs, signatures, etc.).

When a node receives a Lite Block, it looks through the list of transaction hashes the block contains, checks its local transaction pool, and plucks the needed transactions from the pool. If it encounters a transaction that is not in its transaction pool (for any reason), it reaches out to its peers and requests just that transaction. Once the node has all the transactions the block contains, the block is verified, and added to the node’s local copy of the blockchain.

The node then starts the relaying process to propagate the new block to its peers. In going through the list of connected peers, it identifies which of its peers support Lite Blocks and which do not support Lite Blocks. It relays the appropriate block type (Full or Lite) to its peers as needed.

If you’re keen for additional reading on the underlying concept of Lite Blocks check out BTC’s BIP-0152 as well as Monero’s Fluffy Blocks.

TurtleCoin’s Implementation

Adding Lite Block functionality results in significantly smaller blocks propagating across the network. In testing, we’ve observed Lite Blocks as little as 1% of the full block size. Transferring 1KB versus 100KB is a momentous reduction in the amount of data transferred between nodes. This reduction helps to improve block propagation times in a massive way. TurtleCoin nodes will now prefer to send Lite Blocks whenever possible to reduce the size of the information exchanged between nodes. The node will fall back to Full Blocks only when necessary.

@rashedmyt has completed the work required to add Lite Blocks into the core TurtleCoin code. He’s built it in such a way that it does not require a fork of any kind as it relies on the P2P version that we use in a couple of different places. Our plans are to activate Lite Block capability with an upgrade to P2P Version 4. This P2P update will make it into the next release of the core TurtleCoin software.

The addition of Lite Blocks was listed as a TurtleCoin bounty and @rashedmyt will be picking up a 2.05M TRTL bounty for the efforts he’s put forth in getting Lite Blocks implemented in the core code. When you see him in Discord give him a high-five and a round of applause.

Clap

(35)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.