For nearly a decade now, blocks have been relayed through BIP 152 Compact Blocks. This protocol works by sending a a lot smaller abstract of the block and having nodes reconstruct the block utilizing transactions that they’ve already obtained and saved someplace. In fact, if there are transactions that the receiver would not learn about, the receiver should ask the sender for these transactions in order that it will probably full the block and validate it.
A Compact Block is basically the block header, the coinbase transaction, and an inventory of transaction ids of the transactions in that block. It is a far more compact information construction than the complete block itself, so it reduces block propagation delay since much less information must be transmitted for a block on the time it’s mined.
Due to Compact Blocks, massive blocks, and in notably blocks containing massive transactions, shouldn’t have an outsized affect on the time to transmit the block. If a node already has the transactions as a result of it was relayed and saved previous to the block being mined, then the transaction doesn’t should be relayed to the node once more in a block. Moreover, the limiting consider block propagation delay adjustments from the bodily measurement of the block to the variety of transactions within the block, and what number of of these transactions have been already recognized by the receiving node.
Whereas we do not have measurements particularly for blocks containing bigger OP_RETURNs, we do have basic measurements of Compact Block reconstruction charges and reconstruction instances when blocks include non-standard transactions. Reconstruction fee is the share of blocks {that a} node was capable of reconstruct with out asking the sender for lacking transactions. Reconstruction time is the period of time between receiving the Compact Block message, and having the block reconstructed and able to be validated.
There are a number of measurements of Compact Block reconstruction charges and instances over the previous 12 months on this Delving Bitcoin thread: https://delvingbitcoin.org/t/stats-on-compact-block-reconstructions/1052
From the thread, we are able to observe a few totally different durations over the previous 12 months the place blocks have been containing a lot of transactions that weren’t being propagated over the P2P community. We will additionally observe that altering the coverage parameters on some nodes to be extra permissive improves the reconstruction charges and instances.
Particularly, the thread begins with a chart exhibiting that in February 2024, reconstruction charges have been within the 60%-70% vary, dropped right down to 30%-40% for a number of days in March, elevated again to 60%-70%, after which dropped to 10%-20% in April. On the backside of that chart, one of many nodes out of the blue has a 80%-90% reconstruction fee, when the opposite nodes are nonetheless at 60%-70%, and that is correlated with the poster enabling the mempoolfullrbf on that node. That flag made their node extra permissive within the transactions it might settle for to its mempool, which elevated its reconstruction fee.
That preliminary put up additionally has a plot of the reconstruction instances of the blocks, categorized by whether or not a block wanted to request lacking transactions. Blocks that reconstructed with no lacking transactions took about 15ms to reconstruct, whereas blocks that needed to request transactions took about 130ms.
Additional down in that thread, there are some more moderen numbers, though I do not suppose any are from particularly when many individuals have been making OP_RETURNs. Nonetheless, we do see a chart exhibiting reconstruction charges dropping from 60%-70% to twenty%-30% on the finish of July. That is correlated with blocks beginning to include transactions which pay a charge lower than 1 sat/vbyte.
From these measurements, we all know that blocks which include transactions that the receiving nodes already is aware of about reconstruct an order of magnitude sooner. We additionally know that altering mempool coverage to be extra permissive has a constructive impact on reconstruction charges when blocks are already containing transactions that weren’t being relayed beneath the default coverage.
The query then is, what impact on stale blocks does a change on propagation delay have? That is tougher to measure on condition that stale blocks occur comparatively occasionally. Nonetheless, we’re capable of infer some issues from historic information.
On one among my nodes (caveat: this node has not at all times been on-line or had good connectivity, particularly within the early days; additionally it is only one node), which has stale block information ranging from block 340000, there have been a number of stale blocks each week till round 440000 when it decreases to about 1 per week. Round 500000, it drops rapidly to about one stale block each few weeks. Compact Blocks was launched in Bitcoin Core 0.13.0 which was printed in Auguest 2016 (round block 427000), so I believe it might be cheap to imagine that Compact Blocks had a hand in lowering the stale block fee. Others might have extra concrete/dependable numbers exhibiting this, as I disclaim earlier, this is just one node, and one which has not at all times been steady or properly related.
Though not one of the above measurements particularly pertain to blocks containing OP_RETURNs, there isn’t any cause to imagine that OP_RETURNs are a particular case which might trigger the other conduct. On condition that we all know how Compact Blocks works in idea, we are able to observe how the stale block fee has decreased over the previous a number of years and assume that Compact Blocks was a big a part of that. We will observe how Compact Block reconstruction charges have modified over time, and observe that these charges have particularly dropped when miners have been together with non-propagated transactions of their blocks. Contemplating these, an inexpensive conclusion is that enabling transactions with bigger OP_RETURNs to be higher propagated throughout the community would make block propagation sooner, on common, as miners embody bigger OP_RETURNs of their blocks.
