Throughput Statistics from the Steady Lads Mint of June 2022

[gregorycoppola]

Abstract

On June 29, 2022, there was a relatively large NFT mint on the Stacks chain. This was Megapont’s 10k NFT Steady Lads mint.

This is an interesting event because it is a relatively large stress test of the Stacks network.

This discussion looks at some statistics from the launch.

Throughput vs. Mempool Visualization

We are looking at a graph of transactions per block versus mempool size.

There has previously been debate about whether unfilled blocks are only a result of there not being any work to do on them in the mempool. From this graph, we can see that:

• there are many single-digit blocks, even during the middle of the launch, with multi-thousands of transactions in the mempool
• some miners manage to get relatively full blocks of hundreds of transactions, even up to 1000 transactions per block
  • but, at the same time, other miners are producing nearly empty blocks
• the mint of 10k NFT’s took around 3 days to clear

Throughput Statistics

We calculated throughput statistics over the following roughly three-day period, comprising what looked like the bulk of the mint:

• start 2022-06-29T20:29:00.000Z
• end 2022-07-02T16:40:51.000Z

We made statistics for the following measures:

• transactions per second
  • amortized over the past 6 Stacks blocks
  • compares total transactions to time elapsed
• transactions per block
  • this information is part of the block itself
• block fullness
  • the percentage (out of 100%) of the block that was used
    • measured as the percentage of the read count used
    • read count is basically always the bottleneck

We observed the following data:

	transactions / second	transactions / block	percent full
average	0.108	76.5	19.1%
maximum	0.944	1054	100%

Underlying Data

This is based on the following raw data:

• mempool
  • https://raw.githubusercontent.com/gregorycoppola/stx-share/main/data/steady-lads-1-mempool.raw
• blocks
  • https://raw.githubusercontent.com/gregorycoppola/stx-share/main/scrape-blocks/output/2022jul17

[wileyj]

@jcnelson @kantai thoughts?

[wileyj]

@obycode

[obycode]

It looks like this mempool data collected is just a count of the number of transactions in the mempool. It would be great if we collected snapshots of all of the transactions in the mempool, so that experiments could be done to evaluate different methods for selecting and mining.