Skip to main content
This guide provides recommendations for hardware, client software, and configuration settings to optimize the performance of your Base node.

​
Hardware

Running a performant Base node requires adequate hardware. We recommend the following minimum specifications:
  1. A modern multi-core CPU with good single-core performance.
  2. At least 32 GB RAM (64 GB recommended).
  3. A locally attached NVMe SSD drive. RAID 0 configurations can improve performance.
  4. Sufficient storage capacity calculated as:
(2 \* [current chain size](https://base.org/stats) + [snapshot size](https://basechaindata.vercel.app) + 20% buffer)
This accounts for chain data growth and snapshot restoration space.
If utilizing Amazon Elastic Block Store (EBS), io2 Block Express volumes are recommended to ensure sufficient disk read speeds, preventing latency issues during initial sync. However, locally attached NVMe SSDs are strongly recommended over networked storage for optimal performance.

​
Production Hardware Examples

The following are the hardware specifications used for Base production nodes:
  • Reth Archive Node (recommended):
    • Instance: AWS i7i.12xlarge or larger
    • Storage: RAID 0 of all local NVMe drives (/dev/nvme*)
    • Filesystem: ext4
  • Geth Full Node:
    • Instance: AWS i7i.12xlarge or larger
    • Storage: RAID 0 of all local NVMe drives (/dev/nvme*)
    • Filesystem: ext4

​
Initial Sync

Using a recent snapshot can significantly reduce the time required for the initial node synchronization process.

​
Client Software

The Base Node repository contains the current stable configurations and instructions for running different client implementations. Reth is currently the most performant client for running Base nodes. Future optimizations will primarily focus on Reth. You can read more about the migration to Reth here.

​
Geth Performance Tuning (deprecated)

Geth is no longer supported and Reth is the recommended client and shown to be more performant. We recommend migrating Geth nodes to Reth, especially if you are experiencing performance issues.

​
Geth Cache Settings

For Geth nodes, tuning cache allocation via environment variables can improve performance. These settings are used in the standard Docker configuration: 
# .env.mainnet / .env.sepolia
GETH_CACHE="20480"             # Total P2P cache memory allowance (MB) (default: 1024)
GETH_CACHE_DATABASE="20"       # Percentage of cache memory allowance for database IO (default: 75)
GETH_CACHE_GC="12"             # Percentage of cache memory allowance for garbage collection (default: 25)
GETH_CACHE_SNAPSHOT="24"       # Percentage of cache memory allowance for snapshot caching (default: 10)
GETH_CACHE_TRIE="44"           # Percentage of cache memory allowance for trie caching (default: 25)

​
Geth LevelDB Tuning

For teams running Geth with LevelDB, the following patch allows setting LevelDB initialization parameters via environment variables: https://github.com/0x00101010/goleveldb/commit/55ef3429673fb70d389d052a15a4423e13d8b43c This patch can be applied using a replace directive in go.mod when building op-geth. Here’s how to modify your Dockerfile: 
RUN git clone $REPO --branch $VERSION --single-branch . && \
    git switch -c branch-$VERSION $COMMIT && \
    bash -c '[ "$(git rev-parse HEAD)" = "$COMMIT" ]'

RUN echo '' >> go.mod && \
    echo 'replace github.com/syndtr/goleveldb => github.com/0x00101010/goleveldb v1.0.4-param-customization' >> go.mod && \
    go mod tidy

# Continue building op-geth
COPY op-geth/ ./
RUN go run build/ci.go install -static ./cmd/geth
Recommended LevelDB environment variable values with this patch: 
# Recommended LevelDB Settings
LDB_BLOCK_SIZE="524288"              # 512 KiB block size (matches common RAID 0 chunk sizes)
LDB_COMPACTION_TABLE_SIZE="8388608"  # 8 MiB compaction table size (default: 2 MiB)
LDB_COMPACTION_TOTAL_SIZE="41943040" # 40 MiB total compaction size (default: 8 MiB)
LDB_DEBUG_OPTIONS="1"                # Emit LevelDB debug logs