Individual node operators cannot realistically run Base nodes - need better solutions

[ApeachM]

Summary

After extensive testing and research, I've concluded that individual/small-scale operators cannot realistically run a Base full node. I'd like to confirm if my analysis is correct and ask about future plans.

My Testing Environment

• Server: ARM Cortex-X925, 120GB RAM, 3.7TB NVMe SSD
• Goal: Run a Base full node for arbitrage bot RPC

Problems Encountered

1. Archive Snapshot (~8TB)

• Requires 8TB+ SSD (~$1000+ USD)
• Works immediately after download
• Blocker: Storage cost prohibitive for individuals

2. Pruned Snapshot (~1.8TB)

• Fits on affordable SSDs
• Problem: Must execute from block 0 (same as genesis sync)
• Result: 20-30+ days sync time, speed decreases as state grows
• My progress: Started Feb 27, now Mar 16 (17 days), only 82% complete
• Speed dropped from 3,700 blk/min → 450 blk/min (1/8 reduction)

3. HDD Storage

• Tested Archive snapshot on HDD
• Result: ~1.8 blk/min (vs 500+ on SSD)
• 277x slower - completely unusable
• Cannot keep up with new blocks (1 block/2 sec)

4. Snap Sync

• Disabled for OP chains (Enable snap sync #132)
• No timeline for enabling

5. Paid RPC Services

• Rate limited, not unlimited
• Ongoing monthly costs ($49-199+/month)

Questions for Base Team

1. Is my analysis correct? Are these the only options currently available?

2. Pruned snapshot without Execution: Is there any plan to provide a "Full" snapshot that includes receipts/state but excludes historical state? This would allow starting without re-executing from block 0.

3. Snap Sync timeline: When will P2P state sync be enabled for Base?

4. Decentralization concern: If running a node requires 8TB SSD + high-end hardware, how does Base plan to maintain decentralization?

Related Issues

• op-reth Base Mainnet sync very slowly #413 (closed as not planned)
• Enable snap sync #132 (closed but snap sync still unavailable)
• base node slow sync #369 (closed)

Suggestion

Please consider providing a "Full" snapshot option:

• Includes: Current state, receipts
• Excludes: Historical state (account/storage history)
• Size: ~2-3TB (estimate)
• Benefit: No Execution from block 0 required

This would make individual node operation feasible without 8TB storage.

Thank you.