Ethereum News
How Verkle tree testnet-5 activation changes Nethermind stateless sync
The November 2026 testnet-5 activation of Verkle trees modifies how Nethermind handles stateless client sync. This update achieves a 68% witness size reduction. Verkle trees replace Merkle Patricia trees. These structures use polynomial math and KZG commitments instead of simple hashing. A single commitment covers up to 256 children. This flattens the tree compared to hash-based structures. Nethermind uses these smaller proofs to improve sync performance.
Nethermind optimizes its database for high-speed writing during the sync process. The software sets the Sync.TuneDbMode to HeavyWrite by default. This setting helps systems with certain SSD configurations. Users with entry-level NVMe drives can also set this to AggressiveHeavyWrite. This modification provides a boost on specific hardware. Syncing on the Ethereum mainnet involves snap sync, old bodies, old receipts, and state sync. Snap sync downloads the Ethereum state tree. This phase requires high I/O performance. Most SSDs slow down after a few seconds of peak write speed. Therefore, Nethermind relies on high sustained write speeds to finish snap sync. The state sync phase can appear to hang for about 10 minutes as the database compacts for the first time after snap sync. One tested configuration uses an AMD Ryzen 9 7950X CPU and 128GB of RAM. This setup uses Intel Optane SSD 905P Series 900GB storage. A 1Gbps internet connection with TorGuard VPN using the WireGuard protocol provides the network layer. In this configuration, the best test case sync time reaches 1 hour 50 minutes for all phases. The Verkle tree activation via testnet-5 clearly provides the best path for Nethermind’s sync performance.
Archive node migration to Portal Network
Archive node operators moved 23% of their workloads to Portal Network’s distributed storage. This migration follows the Hegota upgrade’s state expiry mechanism. State expiry rolls up old, unused data to prevent state bloat. This prevents the active state from growing indefinitely. Portal Network uses the Blockchain Name System to turn hashes into recognizable names.
The Hegota upgrade combines the Amsterdam execution layer and the Gloas consensus layer. It addresses state bloat through state expiry. This upgrade promotes lower hardware requirements for node operation. This promotes decentralization by allowing more people to run nodes. Verkle trees allow for smaller proofs of data. This means stateless clients can verify transactions without storing the entire chain state. This leads to faster syncing and lower hardware requirements.
| Feature | Merkle Patricia Trie | Verkle Tree |
|---|---|---|
| Witness size (1,000 leaves) | 3.5 MB | 150 kB |
| Math foundation | Hashing | Polynomial commitments |
| Branching factor | Low | Up to 256 |
Akula execution layer rewrite comparison
Akula’s execution layer rewrite presents a different technical direction. This approach competes with the Verkle tree implementation. Nethermind maintains focus on ZK readiness and execution witness capture. The client already supports RISC-V64 compilation. This compatibility helps sequencers generate proofs. You already know that hardware requirements dictate node decentralization. Does the Akula approach provide better hardware compatibility for validators?
Nethermind’s work on ZK readiness includes several completed milestones. The team finished execution witness capture and stateless block replay. They also completed minimal EVM binary creation and RISC-V64 compilation. Nethermind integrates with Zisk, a zkVM environment built on RISC-V64. This integration allows Nethermind to run mainnet blocks through the Zisk pipeline. The shift from Merkle Patricia Tries to Verkle trees utilizes KZG polynomial commitments to group related account data under the same 31-byte stem, which reduces the total number of elements needed for a single proof.
Technical specifications of Hegota
The Hegota upgrade includes EIP-7732. This implements enshrined proposer-builder separation. It separates block proposers from builders. This change separates consensus validation from execution validation. Validators get more time to validate execution payloads. EIP-7928 introduces block-level access lists. These lists record accounts and storage locations accessed during a block. They allow clients to read state from disk in parallel. This allows transactions to process simultaneously.
Because the list contains final values, a freshly started node can adopt the network state without recomputing every transaction again. EIP-8037 changes how Ethereum handles gas for permanent storage. It ties the cost to the volume of data. This sets a fixed price per byte. The goal is a predictable growth of stored data. This helps Ethereum run on commodity hardware. EIP-8037 increases the cost of creating state. It also separates computation and permanent storage gas limits. Developers can roll out larger applications if they fill the storage pot.
