◼️46/100 Block-by-Block: How Ponder optimized its data indexing workflow

One thing I learned about: How Ponder optimized its data indexing workflow

Ponder indexes EVM smart contracts for mobile & web developers.

It works as an ETL pipeline from RPC nodes to a Postgres database.

It aims to process 100,000 events per second.

Three speed optimizations it took:

• Cache events in-memory with periodical flushes in batch
• Move primary key checks to the cache flush step
• Pre-fetch queries based on database access patterns

+ Also tried: static code parser with concurrent indexing

🔽🛠️Resources🔽

• Why Ponder: https://ponder.sh/docs/why-ponder
• Ponder’s (ongoing) journey to 100,000 events per second (2025): https://ponder.sh/blog/performance