Introduction

Polygon is a popular Layer-2 scaling solution that promises fast transaction speed at low fees. It uses the proof-of-stake (PoS) consensus mechanism. PoS validators as block producers create blocks on the Polygon chain at a target/expected block time of 2 seconds. The Merkle root of all blocks produced in between certain intervals are then committed to the Ethereum blockchain for final confirmation.

This dashboard analyses block performance on the Polygon chain, and compares it to the block performance of other Layer 2 and Layer 1 chains. Analysis is on data between 6 June 2022 (the earliest available data) and 24 July 2022.

Definition and calculation

Time between blocks - the time (in seconds) for a block to be created

1. Calculate the time difference between block creations in the polygon.core.fact_blocks schema. The earliest available data is from 6 June 2022.
2. Calculate the average, min, max time between blocks as required.

Transactions per block - the number of transactions validated and included in a block

1. The number of transactions in a block is available in the tx_count field in the polygon.core.fact_blocks schema.
2. Calculate the average number of transactions per block.

Metrics for other L2 and L1 blockchains

The calculation and data time period (i.e. 6 June to 24 July 2022) are the same but data is sourced from the equivalent schemas as follows:

• Arbitrum - arbitrum.core.fact_blocks
• Optimism - optimism.core.fact_blocks
• Ethereum - ethereum.core.fact_blocks
• Solana - solana.core.fact_blocks
• Flow - flow.core.fact_blocks

Block performance overview

• The Polygon chain has a target/expected block time of 2 seconds.
• Between 6 June 2022 and 24 July 2022, Polygon’s average block time was 2.24 seconds, slightly higher than expected. The longest block time was 25 seconds.
• On average, each block contained 74.7 transactions.

Block performance daily trend and correlation

Block time fluctuates over time due to multiple factors. The overall metrics above provides only a high-level trend analysis. The following analysis considers the daily metrics to spot patterns and emerging trends.

• The daily average block time was on an upward trend (i.e. taking longer to produce a block) as shown in the line chart below;
• The average block time was increasing because more blocks were being produced slower than the expected time of 2 seconds. The bar chart below shows the proportion of blocks produced at the expected time; the proportion was above 90% in early June 2022 but dropped to c. 86% in late July 2022.
• Surprisingly, the scatter plot below indicates that there is no correlation between block time and the number of transactions per block. This post on the Polygon Forum suggests that high number of transactions in the mempool (memory pool storing unconfirmed transactions temporarily) could be the cause of the delay in block processing.

Comparison to other blockchains

Average block time

• Polygon is meant to be a Layer 2 scaling solution for Ethereum. In this regard, Polygon is working as intended as it had a much lower block time than Ethereum (2.24 seconds on average vs. 14.3 seconds on average), thus enabling more transactions per second (TPS).
• Other Layer 2 scaling solutions (i.e. Arbitrum and Optimism) appear to be more performant with a lower block time (c. 1 second).
• Flow and Solana are built to have fast transaction speed; it’s not surprising that the average block time is lower than Polygon’s. Solana, in particular, had an average block time of less than 1 second.

Average transactions per block

• Transactions included in a block is a function of usage/demand and the block size (a design choice with trade-offs), hence it’s not a particularly reliable blockchain performance metric.
• For example, Arbitrum and Optimism had an average 1 transaction per block; this low number of transactions per block was likely because of low usage/demand rather than an indication of poor performance. Solana, conversely, had a high number of transactions per block; this does not indicate exceptional performance because validators’ votes (as part of the consensus mechanism) are counted as transactions on Solana.

Conclusion

• Polygon’s block time had been on an upward trend as more blocks were being produced slower than the expected time of 2 seconds.
• The high number of transactions (specifically, unconfirmed transactions on the mempool) could be one of the causes for the delays in producing blocks. This data is not saved on-chain.
• As a Layer 2 scaling solution, Polygon is working as intended as it had a much lower block time than Ethereum (2.24 seconds on average vs. 14.3 seconds on average), thus enabling more transactions per second (TPS).