
Mastering Blockchain Scalability - now on Amazon in all three formats. Born as the textbook for my NTU blockchain course. Rollups, DA, parallel execution, consensus - via Arbitrum, ZKsync, Celestia, Solana, Sui. Amazon: amazon.com/dp/B0HL6PVH2S

Guest lecturer at NTU FinTech; ex-Twilio, PingCAP, SK

Mastering Blockchain Scalability - now on Amazon in all three formats. Born as the textbook for my NTU blockchain course. Rollups, DA, parallel execution, consensus - via Arbitrum, ZKsync, Celestia, Solana, Sui. Amazon: amazon.com/dp/B0HL6PVH2S

Web, PDF, and EPUB are free. Content is CC BY-SA 4.0. Source and corrections are public. If you work on these systems, send one precise correction or missing failure path. github.com/neilydhan/Masteri…

It grew from my blockchain-infrastructure work and SC6019 teaching context at NTU. It is an independent open work, not an institutional endorsement.

The book includes worked calculations, explicit assumptions, graduate exercises and solution sketches, threat-model worksheets, a practitioner evaluation handbook, and a benchmark-reporting template.

Consensus throughput is not execution throughput. Ordering, execution, data propagation, quorum formation, and finality impose different costs. A design can move one bottleneck while leaving the system bound somewhere else.

Parallel execution does not remove shared-state bottlenecks. Independent transactions scale differently from workloads fighting over hot accounts, contracts, queues, or markets. Benchmarks should report the conflict distribution, not only peak throughput.

Data availability is not archival storage. A system can prove that data was available at a decision point without promising that every historical byte remains served forever. The distinction changes both architecture and recovery.

Rollups are not one security model. Ask who sequences, where data goes, who proves, what the bridge verifies, how upgrades work, whether forced inclusion exists, and what users can do when the normal path stops.

The book follows transactions end to end: submission → ordering → execution → data → proof/dispute → settlement → recovery. That same path is used to compare L1s, L2s, channels, rollups, modular systems, and bridges.

A TPS result is incomplete unless you know: * workload and hardware * state and contention * data publication * verification cost * finality rule * trust assumptions * failure and recovery path

I published Mastering Blockchain Scalability, free to read and download. Its argument: scalability is a resource, systems, security, and recovery problem, not a contest for the largest TPS number. neilydhan.github.io/Masterin…

Mastering Blockchain Scalability is out, and free. Most scalability debates end at TPS - the wrong unit. Follows the full path: execution, data availability, consensus, settlement, recovery. L1, L2, rollups, parallel execution. neilydhan.github.io/Masterin…

Kudos to @SimonChin1891 to lead the team to build it!

Meanwhile, we’re exploring new ways to boost adoption via GEO and infra designed for Vibe coding. The goal is to make crypto-native payments just faster, more accessible for developers and businesses. This is just the beginning and I’m excited to share more with you all soon. 🚀

Logically speaking, true innovation is often an idea most people doubt or dismiss — yet it inevitably happens and drives society forward~

AI agent verticals are growing like crazy — even single DB usage is skyrocketing. But only those with the right GTM will capture the wave 😏 Otherwise, you’re just watching from the sidelines.