diff --git a/arc/README.md b/arc/README.md
new file mode 100644
index 000000000..e9e644e03
--- /dev/null
+++ b/arc/README.md
@@ -0,0 +1,175 @@
+# Arc - ClickBench Benchmark
+
+Arc is a high-performance time-series data warehouse built on DuckDB, Parquet, and object storage.
+
+## System Information
+
+- **System:** Arc
+- **Date:** 2025-10-15
+- **Machine:** m3_max (14 cores, 36GB RAM)
+- **Tags:** Python, time-series, DuckDB, Parquet, columnar, HTTP API
+- **License:** AGPL-3.0
+- **Repository:** https://github.com/Basekick-Labs/arc
+
+## Performance
+
+Arc achieves:
+- **Write throughput:** 1.89M records/sec (MessagePack binary protocol)
+- **ClickBench:** ~22 seconds total (43 analytical queries)
+- **Storage:** DuckDB + Parquet with MinIO/S3/GCS backends
+
+## Prerequisites
+
+- Ubuntu/Debian Linux (or compatible)
+- Python 3.11+
+- 8GB+ RAM recommended
+- Internet connection for dataset download
+- Sudo access (only if system dependencies are missing)
+
+## Quick Start
+
+The benchmark script handles everything automatically:
+
+```bash
+./benchmark.sh
+```
+
+This will:
+1. Create Python virtual environment (no system packages modified)
+2. Clone Arc repository
+3. Install dependencies in venv
+4. Start Arc server with optimal worker count (2x CPU cores)
+5. Download ClickBench dataset (14GB parquet file)
+6. Run 43 queries × 3 iterations
+7. Output results in ClickBench JSON format
+
+## Manual Steps
+
+### 1. Install Dependencies
+
+```bash
+sudo apt-get update -y
+sudo apt-get install -y python3-pip python3-venv wget curl
+```
+
+### 2. Create Virtual Environment
+
+```bash
+python3 -m venv arc-venv
+source arc-venv/bin/activate
+```
+
+### 3. Clone and Setup Arc
+
+```bash
+git clone https://github.com/Basekick-Labs/arc.git
+cd arc
+pip install -r requirements.txt
+mkdir -p data logs
+```
+
+### 4. Create API Token
+
+```bash
+python3 << 'EOF'
+from api.auth import AuthManager
+
+auth = AuthManager(db_path='./data/arc.db')
+token = auth.create_token(name='clickbench', description='ClickBench benchmark')
+print(f"Token: {token}")
+EOF
+```
+
+### 5. Start Arc Server
+
+```bash
+# Auto-detect cores
+CORES=$(nproc)
+WORKERS=$((CORES * 2))
+
+# Start server
+gunicorn -w $WORKERS -b 0.0.0.0:8000 \
+    -k uvicorn.workers.UvicornWorker \
+    --timeout 300 \
+    api.main:app
+```
+
+### 6. Download Dataset
+
+```bash
+wget https://datasets.clickhouse.com/hits_compatible/hits.parquet
+```
+
+### 7. Run Benchmark
+
+```bash
+export ARC_URL="http://localhost:8000"
+export ARC_API_KEY="your-token-from-step-4"
+export DATABASE="clickbench"
+export TABLE="hits"
+
+./run.sh
+```
+
+**Note:** The benchmark uses Apache Arrow columnar format for optimal performance. Requires `pyarrow` to be installed.
+
+## Configuration
+
+Arc uses optimal settings for ClickBench (all automatic, no configuration needed):
+
+- **Workers:** Auto-detected cores × 2 (optimal for analytical workloads)
+- **Query cache:** Disabled (per ClickBench rules)
+- **Storage:** Local filesystem (fastest for single-node)
+- **Timeout:** 300 seconds per query
+- **Format:** Apache Arrow (columnar, high-performance)
+
+## Results Format
+
+Results are output in official ClickBench format:
+
+```
+Load time: 0
+Data size: 14779976446
+[0.0226, 0.0233, 0.0284]
+[0.0324, 0.0334, 0.0392]
+...
+```
+
+- **Load time:** Arc queries Parquet files directly without a data loading phase (load time = 0)
+- **Data size:** Size of the dataset in bytes (14GB)
+- **Query results:** 43 lines, each containing 3 execution times (in seconds) for the same query
+
+## Notes
+
+- **Virtual Environment:** All dependencies installed in isolated venv (no `--break-system-packages` needed)
+- **Authentication:** Uses Arc's built-in token auth (simpler than Permission-based auth)
+- **Query Cache:** Disabled to ensure fair benchmark (no cache hits)
+- **Worker Count:** Auto-detected based on CPU cores, optimized for analytical workloads
+- **Timeout:** Generous 300s timeout for complex queries
+
+## Architecture
+
+```
+ClickBench Query → Arc Arrow API → DuckDB → Parquet File → Arrow Results
+```
+
+Arc queries the Parquet file directly via DuckDB's `read_parquet()` function and returns results in Apache Arrow columnar format for maximum efficiency.
+
+## Performance Characteristics
+
+Arc is optimized for:
+- **High-throughput writes** (1.89M RPS with MessagePack)
+- **Analytical queries** (DuckDB's columnar engine)
+- **Columnar data transfer** (Apache Arrow IPC for efficient results)
+- **Object storage** (S3, GCS, MinIO compatibility)
+- **Time-series workloads** (built-in time-based indexing)
+
+## Support
+
+- GitHub: https://github.com/Basekick-Labs/arc
+- Issues: https://github.com/Basekick-Labs/arc/issues
+- Docs: https://docs.arc.basekick.com (coming soon)
+
+## License
+
+Arc Core is licensed under AGPL-3.0.
diff --git a/arc/benchmark.sh b/arc/benchmark.sh
new file mode 100755
index 000000000..55451132e
--- /dev/null
+++ b/arc/benchmark.sh
@@ -0,0 +1,393 @@
+#!/bin/bash
+# Arc ClickBench Complete Benchmark Script
+# This script installs Arc, loads data, and runs the benchmark
+
+set -e
+
+# Check and install system dependencies
+echo "Checking system dependencies..."
+
+MISSING_DEPS=()
+command -v python3 >/dev/null 2>&1 || MISSING_DEPS+=("python3")
+command -v pip3 >/dev/null 2>&1 || MISSING_DEPS+=("python3-pip")
+command -v wget >/dev/null 2>&1 || MISSING_DEPS+=("wget")
+command -v curl >/dev/null 2>&1 || MISSING_DEPS+=("curl")
+
+# Check for python3-venv by detecting Python version
+PYTHON_VERSION=$(python3 --version 2>&1 | awk '{print $2}' | cut -d. -f1,2)
+VENV_PACKAGE="python${PYTHON_VERSION}-venv"
+
+# Try to create a test venv to check if venv is properly installed
+if ! python3 -m venv --help >/dev/null 2>&1 || ! python3 -c "import ensurepip" 2>/dev/null; then
+    MISSING_DEPS+=("$VENV_PACKAGE")
+fi
+
+if [ ${#MISSING_DEPS[@]} -eq 0 ]; then
+    echo "[OK] All system dependencies are already installed"
+else
+    echo "Installing missing dependencies: ${MISSING_DEPS[*]}"
+    sudo apt-get update -y
+    sudo apt-get install -y "${MISSING_DEPS[@]}"
+fi
+
+# Create Python virtual environment
+echo "Creating Python virtual environment..."
+python3 -m venv arc-venv
+source arc-venv/bin/activate
+
+# Clone Arc repository if not exists
+if [ ! -d "arc" ]; then
+    echo "Cloning Arc repository..."
+    git clone https://github.com/Basekick-Labs/arc.git
+fi
+
+cd arc
+
+# Install Arc dependencies in venv
+echo "Installing Arc dependencies..."
+pip install --upgrade pip
+pip install -r requirements.txt
+
+# Create data directory
+mkdir -p data logs
+
+# Create or reuse API token for benchmark
+echo "Setting up API token..."
+python3 << 'EOF'
+from api.auth import AuthManager
+import os
+import time
+
+# Initialize auth manager
+auth = AuthManager(db_path='./data/arc.db')
+
+# Try to create token, or reuse if exists
+token = None
+token_name = f'clickbench-{int(time.time())}'
+
+try:
+    # Try to create new token with timestamp
+    token = auth.create_token(
+        name=token_name,
+        description='ClickBench benchmark access'
+    )
+    print(f"Created new API token: {token_name}")
+except Exception as e:
+    # If that fails, try with a simple name and catch if exists
+    try:
+        token = auth.create_token(
+            name='clickbench',
+            description='ClickBench benchmark access'
+        )
+        print(f"Created API token: clickbench")
+    except ValueError:
+        # Token already exists, list and use existing one
+        print("Token 'clickbench' already exists, retrieving...")
+        tokens = auth.list_tokens()
+        for t in tokens:
+            if t.get('name') == 'clickbench':
+                token = t.get('token')
+                print(f"Reusing existing token: clickbench")
+                break
+
+        if not token:
+            raise Exception("Could not create or retrieve token")
+
+# Write token to file for run.sh to use
+with open('../arc_token.txt', 'w') as f:
+    f.write(token)
+EOF
+
+ARC_TOKEN=$(cat ../arc_token.txt)
+echo "Token ready: $ARC_TOKEN"
+
+# Auto-detect CPU cores (supports Linux and macOS)
+if command -v nproc > /dev/null 2>&1; then
+    # Linux: use nproc
+    CORES=$(nproc)
+elif command -v sysctl > /dev/null 2>&1; then
+    # macOS: use sysctl
+    CORES=$(sysctl -n hw.ncpu 2>/dev/null || sysctl -n hw.logicalcpu 2>/dev/null || echo 4)
+elif [ -f /proc/cpuinfo ]; then
+    # Linux fallback: parse /proc/cpuinfo
+    CORES=$(grep -c processor /proc/cpuinfo)
+else
+    # Final fallback
+    CORES=4
+fi
+
+# Use 2x cores for optimal analytical performance (automatic)
+WORKERS=$((CORES * 2))
+echo "Starting Arc with $WORKERS workers ($CORES cores detected, 2x multiplier for optimal performance)..."
+
+# Create minimal .env if not exists
+if [ ! -f ".env" ]; then
+    cat > .env << 'ENVEOF'
+# Arc Configuration for ClickBench
+STORAGE_BACKEND=local
+LOCAL_STORAGE_PATH=./minio-data
+PORT=8000
+HOST=0.0.0.0
+LOG_LEVEL=WARNING
+QUERY_CACHE_ENABLED=false
+BUFFER_MAX_SIZE=50000
+BUFFER_MAX_AGE=5
+ENVEOF
+fi
+
+# Start Arc server in background
+gunicorn -w $WORKERS -b 0.0.0.0:8000 \
+    -k uvicorn.workers.UvicornWorker \
+    --timeout 300 \
+    --access-logfile /dev/null \
+    --error-logfile ../arc.log \
+    --log-level warning \
+    api.main:app > /dev/null 2>&1 &
+
+ARC_PID=$!
+echo "Arc started with PID: $ARC_PID"
+
+# Wait for Arc to be ready (up to 30 seconds)
+echo "Waiting for Arc to be ready..."
+for i in {1..30}; do
+    if curl -s -f http://localhost:8000/health > /dev/null 2>&1; then
+        echo "[OK] Arc is ready!"
+        break
+    fi
+    if [ $i -eq 30 ]; then
+        echo "Error: Arc failed to start within 30 seconds"
+        echo "Last 50 lines of logs:"
+        tail -50 ../arc.log
+        kill $ARC_PID 2>/dev/null || true
+        exit 1
+    fi
+    sleep 1
+done
+
+cd ..
+
+# Download and prepare dataset
+DATASET_FILE="hits.parquet"
+DATASET_URL="https://datasets.clickhouse.com/hits_compatible/hits.parquet"
+EXPECTED_SIZE=14779976446  # 14GB
+
+if [ -f "$DATASET_FILE" ]; then
+    CURRENT_SIZE=$(stat -f%z "$DATASET_FILE" 2>/dev/null || stat -c%s "$DATASET_FILE" 2>/dev/null)
+    if [ "$CURRENT_SIZE" -eq "$EXPECTED_SIZE" ]; then
+        echo "[OK] Dataset already downloaded (14GB)"
+    else
+        echo "[WARNING] Dataset exists but size mismatch (expected: $EXPECTED_SIZE, got: $CURRENT_SIZE)"
+        echo "Re-downloading dataset..."
+        rm -f "$DATASET_FILE"
+        wget --continue --progress=dot:giga "$DATASET_URL"
+    fi
+else
+    echo "Downloading ClickBench dataset (14GB)..."
+    wget --continue --progress=dot:giga "$DATASET_URL"
+fi
+
+FILE_SIZE=$(du -h "$DATASET_FILE" | cut -f1)
+echo "Dataset size: $FILE_SIZE ($DATASET_FILE)"
+
+# Count rows using DuckDB
+echo "Counting rows..."
+python3 << 'EOF'
+import duckdb
+conn = duckdb.connect()
+count = conn.execute("SELECT COUNT(*) FROM read_parquet('hits.parquet')").fetchone()[0]
+print(f"Dataset contains {count:,} rows")
+EOF
+
+# Set environment variables for benchmarking
+export ARC_URL="http://localhost:8000"
+export ARC_API_KEY="$ARC_TOKEN"
+export DATABASE="clickbench"
+export TABLE="hits"
+
+# Load data into Arc by copying parquet file to storage
+echo ""
+echo "Loading ClickBench data into Arc..."
+echo "================================================"
+
+STORAGE_BASE="arc/data/arc"
+TARGET_DIR="$STORAGE_BASE/$DATABASE/$TABLE"
+TARGET_FILE="$TARGET_DIR/hits.parquet"
+
+# Create target directory
+mkdir -p "$TARGET_DIR"
+
+# Check if already loaded
+if [ -f "$TARGET_FILE" ]; then
+    SOURCE_SIZE=$(stat -f%z "$DATASET_FILE" 2>/dev/null || stat -c%s "$DATASET_FILE" 2>/dev/null)
+    TARGET_SIZE=$(stat -f%z "$TARGET_FILE" 2>/dev/null || stat -c%s "$TARGET_FILE" 2>/dev/null)
+
+    if [ "$SOURCE_SIZE" -eq "$TARGET_SIZE" ]; then
+        echo "[OK] Data already loaded (14GB)"
+        echo "  Location: $TARGET_FILE"
+    else
+        echo "[WARNING] Existing file has different size, reloading..."
+        rm -f "$TARGET_FILE"
+        echo "  Copying parquet file to Arc storage..."
+        cp "$DATASET_FILE" "$TARGET_FILE"
+        echo "[OK] Data loaded successfully!"
+    fi
+else
+    echo "  Copying parquet file to Arc storage..."
+    echo "  Source: $DATASET_FILE"
+    echo "  Target: $TARGET_FILE"
+    cp "$DATASET_FILE" "$TARGET_FILE"
+    echo "[OK] Data loaded successfully!"
+    echo "  Table: $DATABASE.$TABLE"
+    TARGET_SIZE=$(du -h "$TARGET_FILE" | cut -f1)
+    echo "  Size: $TARGET_SIZE"
+fi
+
+echo ""
+echo "Data loading complete."
+
+# Verify query cache configuration
+echo ""
+echo "Verifying query cache configuration..."
+cd arc
+python3 << 'CACHECHECK'
+import os
+import sys
+
+# Check all possible cache configuration sources
+print("=" * 70)
+print("Query Cache Configuration Check")
+print("=" * 70)
+
+# 1. Check arc.conf
+cache_in_conf = None
+try:
+    from config_loader import get_config
+    arc_config = get_config()
+    cache_config = arc_config.config.get('query_cache', {})
+    cache_in_conf = cache_config.get('enabled', None)
+    print(f"  arc.conf:     enabled = {cache_in_conf}")
+except Exception as e:
+    print(f"  arc.conf:     Error reading: {e}")
+
+# 2. Check .env file
+cache_in_env = None
+if os.path.exists('.env'):
+    with open('.env', 'r') as f:
+        for line in f:
+            if line.strip().startswith('QUERY_CACHE_ENABLED'):
+                cache_in_env = line.split('=')[1].strip().lower()
+                print(f"  .env:         QUERY_CACHE_ENABLED = {cache_in_env}")
+                break
+    if cache_in_env is None:
+        print(f"  .env:         QUERY_CACHE_ENABLED not set")
+else:
+    print(f"  .env:         File not found")
+
+# 3. Check environment variable
+cache_in_os_env = os.getenv("QUERY_CACHE_ENABLED")
+if cache_in_os_env:
+    print(f"  Environment:  QUERY_CACHE_ENABLED = {cache_in_os_env}")
+else:
+    print(f"  Environment:  QUERY_CACHE_ENABLED not set")
+
+# 4. Check what init_query_cache will actually use
+print("")
+try:
+    from api.query_cache import init_query_cache
+    cache_instance = init_query_cache()
+    if cache_instance is None:
+        print(f"[OK] FINAL RESULT: Query cache is DISABLED")
+    else:
+        print(f"[ERROR] FINAL RESULT: Query cache is ENABLED")
+        print(f"    TTL: {cache_instance.ttl_seconds}s")
+        print(f"    Max size: {cache_instance.max_size}")
+        print(f"\n  [WARNING] Cache must be disabled for valid benchmark results!")
+except Exception as e:
+    print(f"[ERROR] Error checking cache initialization: {e}")
+
+print("=" * 70)
+CACHECHECK
+
+cd ..
+
+# Test API token before running benchmark
+echo ""
+echo "Testing API token authentication..."
+TEST_RESPONSE=$(curl -s -w "\n%{http_code}" -H "x-api-key: $ARC_API_KEY" "$ARC_URL/health")
+HTTP_CODE=$(echo "$TEST_RESPONSE" | tail -n1)
+if [ "$HTTP_CODE" = "200" ]; then
+    echo "[OK] API token is valid"
+else
+    echo "[ERROR] API token test failed (HTTP $HTTP_CODE)"
+    echo "Response: $(echo "$TEST_RESPONSE" | head -n-1)"
+    echo ""
+    echo "Debugging: Let's verify the token exists in the database..."
+    cd arc
+    python3 << 'DEBUGEOF'
+from api.auth import AuthManager
+auth = AuthManager(db_path='./data/arc.db')
+tokens = auth.list_tokens()
+print(f"Found {len(tokens)} tokens in database:")
+for t in tokens:
+    print(f"  - {t.get('name')}: {t.get('token')[:20]}...")
+DEBUGEOF
+    cd ..
+    echo ""
+    echo "Error: Cannot proceed without valid authentication"
+    kill $ARC_PID 2>/dev/null || true
+    exit 1
+fi
+
+# Run benchmark
+echo ""
+echo "Running ClickBench queries via Arc Arrow API..."
+echo "================================================"
+echo "(Logging to log.txt, this may take a few minutes...)"
+./run.sh > log.txt 2>&1
+echo "Benchmark execution complete!"
+
+# Stop Arc
+echo ""
+echo "Stopping Arc..."
+kill $ARC_PID 2>/dev/null || true
+wait $ARC_PID 2>/dev/null || true
+
+# Deactivate venv
+deactivate
+
+# Format results for ClickBench (official format)
+echo ""
+echo "Formatting results..."
+
+# Extract timing values from log
+cat log.txt | grep -oE '^[0-9]+\.[0-9]+|^null' | \
+  awk '{
+    if (NR % 3 == 1) printf "[";
+    printf "%s", $1;
+    if (NR % 3 == 0) print "]";
+    else printf ", ";
+  }' > results.txt
+
+# Output in official ClickBench format
+echo ""
+echo "[OK] Benchmark complete!"
+echo ""
+echo "================================================"
+echo "Official ClickBench Results"
+echo "================================================"
+echo ""
+
+# Load time (Arc doesn't load data, it queries Parquet directly)
+echo "Load time: 0"
+
+# Data size in bytes
+echo "Data size: $EXPECTED_SIZE"
+
+# Query results (43 lines)
+cat results.txt
+
+echo ""
+echo "================================================"
+echo "Results saved to: results.txt"
+echo "Full logs saved to: log.txt"
+echo "================================================"
diff --git a/arc/create.sql b/arc/create.sql
new file mode 100644
index 000000000..9c5f4857a
--- /dev/null
+++ b/arc/create.sql
@@ -0,0 +1,116 @@
+-- Arc ClickBench Schema
+--
+-- Note: Arc queries Parquet files directly via DuckDB, so no explicit table creation is needed.
+-- The benchmark.sh script copies the hits.parquet file to Arc's storage directory:
+--   arc/data/arc/clickbench/hits/hits.parquet
+--
+-- Arc automatically detects and queries Parquet files without requiring schema definition.
+-- This file documents the equivalent schema for reference.
+
+CREATE TABLE hits (
+    WatchID BIGINT,
+    JavaEnable SMALLINT,
+    Title VARCHAR,
+    GoodEvent SMALLINT,
+    EventTime BIGINT,
+    EventDate USMALLINT,
+    CounterID INTEGER,
+    ClientIP INTEGER,
+    RegionID INTEGER,
+    UserID BIGINT,
+    CounterClass SMALLINT,
+    OS SMALLINT,
+    UserAgent SMALLINT,
+    URL VARCHAR,
+    Referer VARCHAR,
+    IsRefresh SMALLINT,
+    RefererCategoryID SMALLINT,
+    RefererRegionID INTEGER,
+    URLCategoryID SMALLINT,
+    URLRegionID INTEGER,
+    ResolutionWidth SMALLINT,
+    ResolutionHeight SMALLINT,
+    ResolutionDepth SMALLINT,
+    FlashMajor SMALLINT,
+    FlashMinor SMALLINT,
+    FlashMinor2 VARCHAR,
+    NetMajor SMALLINT,
+    NetMinor SMALLINT,
+    UserAgentMajor SMALLINT,
+    UserAgentMinor VARCHAR,
+    CookieEnable SMALLINT,
+    JavascriptEnable SMALLINT,
+    IsMobile SMALLINT,
+    MobilePhone SMALLINT,
+    MobilePhoneModel VARCHAR,
+    Params VARCHAR,
+    IPNetworkID INTEGER,
+    TraficSourceID SMALLINT,
+    SearchEngineID SMALLINT,
+    SearchPhrase VARCHAR,
+    AdvEngineID SMALLINT,
+    IsArtifical SMALLINT,
+    WindowClientWidth SMALLINT,
+    WindowClientHeight SMALLINT,
+    ClientTimeZone SMALLINT,
+    ClientEventTime BIGINT,
+    SilverlightVersion1 SMALLINT,
+    SilverlightVersion2 SMALLINT,
+    SilverlightVersion3 INTEGER,
+    SilverlightVersion4 SMALLINT,
+    PageCharset VARCHAR,
+    CodeVersion INTEGER,
+    IsLink SMALLINT,
+    IsDownload SMALLINT,
+    IsNotBounce SMALLINT,
+    FUniqID BIGINT,
+    OriginalURL VARCHAR,
+    HID INTEGER,
+    IsOldCounter SMALLINT,
+    IsEvent SMALLINT,
+    IsParameter SMALLINT,
+    DontCountHits SMALLINT,
+    WithHash SMALLINT,
+    HitColor VARCHAR,
+    LocalEventTime BIGINT,
+    Age SMALLINT,
+    Sex SMALLINT,
+    Income SMALLINT,
+    Interests SMALLINT,
+    Robotness SMALLINT,
+    RemoteIP INTEGER,
+    WindowName INTEGER,
+    OpenerName INTEGER,
+    HistoryLength SMALLINT,
+    BrowserLanguage VARCHAR,
+    BrowserCountry VARCHAR,
+    SocialNetwork VARCHAR,
+    SocialAction VARCHAR,
+    HTTPError SMALLINT,
+    SendTiming INTEGER,
+    DNSTiming INTEGER,
+    ConnectTiming INTEGER,
+    ResponseStartTiming INTEGER,
+    ResponseEndTiming INTEGER,
+    FetchTiming INTEGER,
+    SocialSourceNetworkID SMALLINT,
+    SocialSourcePage VARCHAR,
+    ParamPrice BIGINT,
+    ParamOrderID VARCHAR,
+    ParamCurrency VARCHAR,
+    ParamCurrencyID SMALLINT,
+    OpenstatServiceName VARCHAR,
+    OpenstatCampaignID VARCHAR,
+    OpenstatAdID VARCHAR,
+    OpenstatSourceID VARCHAR,
+    UTMSource VARCHAR,
+    UTMMedium VARCHAR,
+    UTMCampaign VARCHAR,
+    UTMContent VARCHAR,
+    UTMTerm VARCHAR,
+    FromTag VARCHAR,
+    HasGCLID SMALLINT,
+    RefererHash BIGINT,
+    URLHash BIGINT,
+    CLID INTEGER
+);
diff --git a/arc/queries.sql b/arc/queries.sql
new file mode 100644
index 000000000..64b056b91
--- /dev/null
+++ b/arc/queries.sql
@@ -0,0 +1,47 @@
+-- ClickBench Queries for Arc
+-- Original: https://github.com/ClickHouse/ClickBench
+-- Adapted for Arc's DuckDB engine (table name: clickbench.hits)
+
+SELECT COUNT(*) FROM clickbench.hits;
+SELECT COUNT(*) FROM clickbench.hits WHERE AdvEngineID <> 0;
+SELECT SUM(AdvEngineID), COUNT(*), AVG(ResolutionWidth) FROM clickbench.hits;
+SELECT AVG(UserID) FROM clickbench.hits;
+SELECT COUNT(DISTINCT UserID) FROM clickbench.hits;
+SELECT COUNT(DISTINCT SearchPhrase) FROM clickbench.hits;
+SELECT MIN(EventDate), MAX(EventDate) FROM clickbench.hits;
+SELECT AdvEngineID, COUNT(*) FROM clickbench.hits WHERE AdvEngineID <> 0 GROUP BY AdvEngineID ORDER BY COUNT(*) DESC;
+SELECT RegionID, COUNT(DISTINCT UserID) AS u FROM clickbench.hits GROUP BY RegionID ORDER BY u DESC LIMIT 10;
+SELECT RegionID, SUM(AdvEngineID), COUNT(*) AS c, AVG(ResolutionWidth), COUNT(DISTINCT UserID) FROM clickbench.hits GROUP BY RegionID ORDER BY c DESC LIMIT 10;
+SELECT MobilePhoneModel, COUNT(DISTINCT UserID) AS u FROM clickbench.hits WHERE MobilePhoneModel <> '' GROUP BY MobilePhoneModel ORDER BY u DESC LIMIT 10;
+SELECT MobilePhone, MobilePhoneModel, COUNT(DISTINCT UserID) AS u FROM clickbench.hits WHERE MobilePhoneModel <> '' GROUP BY MobilePhone, MobilePhoneModel ORDER BY u DESC LIMIT 10;
+SELECT SearchPhrase, COUNT(*) AS c FROM clickbench.hits WHERE SearchPhrase <> '' GROUP BY SearchPhrase ORDER BY c DESC LIMIT 10;
+SELECT SearchPhrase, COUNT(DISTINCT UserID) AS u FROM clickbench.hits WHERE SearchPhrase <> '' GROUP BY SearchPhrase ORDER BY u DESC LIMIT 10;
+SELECT SearchEngineID, SearchPhrase, COUNT(*) AS c FROM clickbench.hits WHERE SearchPhrase <> '' GROUP BY SearchEngineID, SearchPhrase ORDER BY c DESC LIMIT 10;
+SELECT UserID, COUNT(*) FROM clickbench.hits GROUP BY UserID ORDER BY COUNT(*) DESC LIMIT 10;
+SELECT UserID, SearchPhrase, COUNT(*) FROM clickbench.hits GROUP BY UserID, SearchPhrase ORDER BY COUNT(*) DESC LIMIT 10;
+SELECT UserID, SearchPhrase, COUNT(*) FROM clickbench.hits GROUP BY UserID, SearchPhrase LIMIT 10;
+SELECT UserID, minute(to_timestamp(EventTime)) AS m, SearchPhrase, COUNT(*) FROM clickbench.hits GROUP BY UserID, m, SearchPhrase ORDER BY COUNT(*) DESC LIMIT 10;
+SELECT UserID FROM clickbench.hits WHERE UserID = 435090932899640449;
+SELECT COUNT(*) FROM clickbench.hits WHERE URL LIKE '%google%';
+SELECT SearchPhrase, MIN(URL), COUNT(*) AS c FROM clickbench.hits WHERE URL LIKE '%google%' AND SearchPhrase <> '' GROUP BY SearchPhrase ORDER BY c DESC LIMIT 10;
+SELECT SearchPhrase, MIN(URL), MIN(Title), COUNT(*) AS c, COUNT(DISTINCT UserID) FROM clickbench.hits WHERE Title LIKE '%Google%' AND URL NOT LIKE '%.google.%' AND SearchPhrase <> '' GROUP BY SearchPhrase ORDER BY c DESC LIMIT 10;
+SELECT * FROM clickbench.hits WHERE URL LIKE '%google%' ORDER BY EventTime LIMIT 10;
+SELECT SearchPhrase FROM clickbench.hits WHERE SearchPhrase <> '' ORDER BY EventTime LIMIT 10;
+SELECT SearchPhrase FROM clickbench.hits WHERE SearchPhrase <> '' ORDER BY SearchPhrase LIMIT 10;
+SELECT SearchPhrase FROM clickbench.hits WHERE SearchPhrase <> '' ORDER BY EventTime, SearchPhrase LIMIT 10;
+SELECT CounterID, AVG(length(URL)) AS l, COUNT(*) AS c FROM clickbench.hits WHERE URL <> '' GROUP BY CounterID HAVING COUNT(*) > 100000 ORDER BY l DESC LIMIT 25;
+SELECT REGEXP_REPLACE(Referer, '^https?://(?:www\.)?([^/]+)/.*$', '\1') AS k, AVG(length(Referer)) AS l, COUNT(*) AS c, MIN(Referer) FROM clickbench.hits WHERE Referer <> '' GROUP BY k HAVING COUNT(*) > 100000 ORDER BY l DESC LIMIT 25;
+SELECT SUM(ResolutionWidth), SUM(ResolutionWidth + 1), SUM(ResolutionWidth + 2), SUM(ResolutionWidth + 3), SUM(ResolutionWidth + 4), SUM(ResolutionWidth + 5), SUM(ResolutionWidth + 6), SUM(ResolutionWidth + 7), SUM(ResolutionWidth + 8), SUM(ResolutionWidth + 9), SUM(ResolutionWidth + 10), SUM(ResolutionWidth + 11), SUM(ResolutionWidth + 12), SUM(ResolutionWidth + 13), SUM(ResolutionWidth + 14), SUM(ResolutionWidth + 15), SUM(ResolutionWidth + 16), SUM(ResolutionWidth + 17), SUM(ResolutionWidth + 18), SUM(ResolutionWidth + 19), SUM(ResolutionWidth + 20), SUM(ResolutionWidth + 21), SUM(ResolutionWidth + 22), SUM(ResolutionWidth + 23), SUM(ResolutionWidth + 24), SUM(ResolutionWidth + 25), SUM(ResolutionWidth + 26), SUM(ResolutionWidth + 27), SUM(ResolutionWidth + 28), SUM(ResolutionWidth + 29), SUM(ResolutionWidth + 30), SUM(ResolutionWidth + 31), SUM(ResolutionWidth + 32), SUM(ResolutionWidth + 33), SUM(ResolutionWidth + 34), SUM(ResolutionWidth + 35), SUM(ResolutionWidth + 36), SUM(ResolutionWidth + 37), SUM(ResolutionWidth + 38), SUM(ResolutionWidth + 39), SUM(ResolutionWidth + 40), SUM(ResolutionWidth + 41), SUM(ResolutionWidth + 42), SUM(ResolutionWidth + 43), SUM(ResolutionWidth + 44), SUM(ResolutionWidth + 45), SUM(ResolutionWidth + 46), SUM(ResolutionWidth + 47), SUM(ResolutionWidth + 48), SUM(ResolutionWidth + 49), SUM(ResolutionWidth + 50), SUM(ResolutionWidth + 51), SUM(ResolutionWidth + 52), SUM(ResolutionWidth + 53), SUM(ResolutionWidth + 54), SUM(ResolutionWidth + 55), SUM(ResolutionWidth + 56), SUM(ResolutionWidth + 57), SUM(ResolutionWidth + 58), SUM(ResolutionWidth + 59), SUM(ResolutionWidth + 60), SUM(ResolutionWidth + 61), SUM(ResolutionWidth + 62), SUM(ResolutionWidth + 63), SUM(ResolutionWidth + 64), SUM(ResolutionWidth + 65), SUM(ResolutionWidth + 66), SUM(ResolutionWidth + 67), SUM(ResolutionWidth + 68), SUM(ResolutionWidth + 69), SUM(ResolutionWidth + 70), SUM(ResolutionWidth + 71), SUM(ResolutionWidth + 72), SUM(ResolutionWidth + 73), SUM(ResolutionWidth + 74), SUM(ResolutionWidth + 75), SUM(ResolutionWidth + 76), SUM(ResolutionWidth + 77), SUM(ResolutionWidth + 78), SUM(ResolutionWidth + 79), SUM(ResolutionWidth + 80), SUM(ResolutionWidth + 81), SUM(ResolutionWidth + 82), SUM(ResolutionWidth + 83), SUM(ResolutionWidth + 84), SUM(ResolutionWidth + 85), SUM(ResolutionWidth + 86), SUM(ResolutionWidth + 87), SUM(ResolutionWidth + 88), SUM(ResolutionWidth + 89) FROM clickbench.hits;
+SELECT SearchEngineID, ClientIP, COUNT(*) AS c, SUM(IsRefresh), AVG(ResolutionWidth) FROM clickbench.hits WHERE SearchPhrase <> '' GROUP BY SearchEngineID, ClientIP ORDER BY c DESC LIMIT 10;
+SELECT WatchID, ClientIP, COUNT(*) AS c, SUM(IsRefresh), AVG(ResolutionWidth) FROM clickbench.hits WHERE SearchPhrase <> '' GROUP BY WatchID, ClientIP ORDER BY c DESC LIMIT 10;
+SELECT WatchID, ClientIP, COUNT(*) AS c, SUM(IsRefresh), AVG(ResolutionWidth) FROM clickbench.hits GROUP BY WatchID, ClientIP ORDER BY c DESC LIMIT 10;
+SELECT URL, COUNT(*) AS c FROM clickbench.hits GROUP BY URL ORDER BY c DESC LIMIT 10;
+SELECT 1, URL, COUNT(*) AS c FROM clickbench.hits GROUP BY 1, URL ORDER BY c DESC LIMIT 10;
+SELECT ClientIP, ClientIP - 1, ClientIP - 2, ClientIP - 3, COUNT(*) AS c FROM clickbench.hits GROUP BY ClientIP, ClientIP - 1, ClientIP - 2, ClientIP - 3 ORDER BY c DESC LIMIT 10;
+SELECT URL, COUNT(*) AS PageViews FROM clickbench.hits WHERE CounterID = 62 AND EventDate >= 15887 AND EventDate <= 15917 AND DontCountHits = 0 AND IsRefresh = 0 AND URL <> '' GROUP BY URL ORDER BY PageViews DESC LIMIT 10;
+SELECT Title, COUNT(*) AS PageViews FROM clickbench.hits WHERE CounterID = 62 AND EventDate >= 15887 AND EventDate <= 15917 AND DontCountHits = 0 AND IsRefresh = 0 AND Title <> '' GROUP BY Title ORDER BY PageViews DESC LIMIT 10;
+SELECT URL, COUNT(*) AS PageViews FROM clickbench.hits WHERE CounterID = 62 AND EventDate >= 15887 AND EventDate <= 15917 AND IsRefresh = 0 AND IsLink <> 0 AND IsDownload = 0 GROUP BY URL ORDER BY PageViews DESC LIMIT 10 OFFSET 1000;
+SELECT TraficSourceID, SearchEngineID, AdvEngineID, CASE WHEN (SearchEngineID = 0 AND AdvEngineID = 0) THEN Referer ELSE '' END AS Src, URL AS Dst, COUNT(*) AS PageViews FROM clickbench.hits WHERE CounterID = 62 AND EventDate >= 15887 AND EventDate <= 15917 AND IsRefresh = 0 GROUP BY TraficSourceID, SearchEngineID, AdvEngineID, Src, Dst ORDER BY PageViews DESC LIMIT 10 OFFSET 1000;
+SELECT URLHash, EventDate, COUNT(*) AS PageViews FROM clickbench.hits WHERE CounterID = 62 AND EventDate >= 15887 AND EventDate <= 15917 AND IsRefresh = 0 AND TraficSourceID IN (-1, 6) AND RefererHash = 3594120000172545465 GROUP BY URLHash, EventDate ORDER BY PageViews DESC LIMIT 10 OFFSET 100;
+SELECT WindowClientWidth, WindowClientHeight, COUNT(*) AS PageViews FROM clickbench.hits WHERE CounterID = 62 AND EventDate >= 15887 AND EventDate <= 15917 AND IsRefresh = 0 AND DontCountHits = 0 AND URLHash = 2868770270353813622 GROUP BY WindowClientWidth, WindowClientHeight ORDER BY PageViews DESC LIMIT 10 OFFSET 10000;
+SELECT DATE_TRUNC('minute', to_timestamp(EventTime)) AS M, COUNT(*) AS PageViews FROM clickbench.hits WHERE CounterID = 62 AND EventDate >= 15900 AND EventDate <= 15901 AND IsRefresh = 0 AND DontCountHits = 0 GROUP BY DATE_TRUNC('minute', to_timestamp(EventTime)) ORDER BY DATE_TRUNC('minute', to_timestamp(EventTime)) LIMIT 10 OFFSET 1000;
diff --git a/arc/results/c6a.4xlarge.json b/arc/results/c6a.4xlarge.json
new file mode 100644
index 000000000..0075400bc
--- /dev/null
+++ b/arc/results/c6a.4xlarge.json
@@ -0,0 +1,56 @@
+{
+  "system": "Arc",
+  "date": "2025-10-16",
+  "machine": "c6a.4xlarge",
+  "cluster_size": 1,
+  "proprietary": "no",
+  "tuned": "no",
+  "tags": ["time-series"],
+  "load_time": 0,
+  "data_size": 14779976446,
+  "result": [
+    [0.1771, 0.0489, 0.0306],
+    [0.0724, 0.0483, 0.0482],
+    [0.0792, 0.0728, 0.0732],
+    [0.0755, 0.0738, 0.0721],
+    [0.3664, 0.3545, 0.3401],
+    [0.5707, 0.5549, 0.5394],
+    [0.0960, 0.0587, 0.0592],
+    [0.1152, 0.0580, 0.0515],
+    [0.4881, 0.4437, 0.4640],
+    [0.6132, 0.5848, 0.5753],
+    [0.1467, 0.1459, 0.1451],
+    [0.1950, 0.1970, 0.1816],
+    [0.5801, 0.5568, 0.5706],
+    [0.9320, 0.9209, 0.8731],
+    [0.5911, 0.5856, 0.5947],
+    [0.4006, 0.4228, 0.4202],
+    [1.0630, 1.0253, 1.0050],
+    [0.7713, 0.7968, 0.7761],
+    [3.3695, 3.3090, 3.3101],
+    [0.0750, 0.0619, 0.0677],
+    [1.0353, 0.9025, 0.9012],
+    [0.8437, 0.8448, 0.8360],
+    [1.6622, 1.6610, 1.6706],
+    [0.5193, 0.5336, 0.5057],
+    [0.1918, 0.1977, 0.1965],
+    [0.2946, 0.2899, 0.2886],
+    [0.1373, 0.1341, 0.1675],
+    [1.0057, 0.9827, 1.0277],
+    [9.0295, 9.1074, 9.1164],
+    [0.0816, 0.0762, 0.0802],
+    [0.7967, 0.6669, 0.5835],
+    [0.6877, 0.6892, 0.6830],
+    [2.0258, 1.9906, 1.9201],
+    [2.2859, 2.3226, 2.3100],
+    [2.4650, 2.4027, 2.4308],
+    [0.5759, 0.7953, 0.6377],
+    [0.2055, 0.1400, 0.1667],
+    [0.1341, 0.1267, 0.1235],
+    [0.0912, 0.0903, 0.0901],
+    [0.2624, 0.2692, 0.2877],
+    [0.0640, 0.0624, 0.0609],
+    [0.0729, 0.0595, 0.0571],
+    [0.2177, 0.2407, 0.2197]
+  ]
+}
diff --git a/arc/results/m3_max.json b/arc/results/m3_max.json
new file mode 100644
index 000000000..d4cab0132
--- /dev/null
+++ b/arc/results/m3_max.json
@@ -0,0 +1,56 @@
+{
+  "system": "Arc",
+  "date": "2025-10-15",
+  "machine": "M3 Pro Max: 14 Cores, 36GB",
+  "cluster_size": 1,
+  "proprietary": "no",
+  "tuned": "no",
+  "tags": ["time-series"],
+  "load_time": 0,
+  "data_size": 14779976446,
+  "result": [
+    [0.0268, 0.0238, 0.0219],
+    [0.0365, 0.0335, 0.0343],
+    [0.0559, 0.0478, 0.0457],
+    [0.0683, 0.0554, 0.0520],
+    [0.1714, 0.1749, 0.1791],
+    [0.2990, 0.2767, 0.2812],
+    [0.0430, 0.0396, 0.0365],
+    [0.0393, 0.0393, 0.0340],
+    [0.2293, 0.2175, 0.2173],
+    [0.3047, 0.3122, 0.3065],
+    [0.0989, 0.0912, 0.0978],
+    [0.1058, 0.1036, 0.1046],
+    [0.2837, 0.2848, 0.3015],
+    [0.4204, 0.4309, 0.4429],
+    [0.3361, 0.3174, 0.3112],
+    [0.2122, 0.2008, 0.2017],
+    [0.5108, 0.5335, 0.5193],
+    [0.4918, 0.4830, 0.4810],
+    [1.6410, 1.6183, 1.6337],
+    [0.0685, 0.0600, 0.0661],
+    [0.8242, 0.5983, 0.5881],
+    [0.4983, 0.4987, 0.4998],
+    [1.0744, 1.0914, 1.1013],
+    [0.3505, 0.3212, 0.3235],
+    [0.1281, 0.1211, 0.1214],
+    [0.1995, 0.1768, 0.1786],
+    [0.0870, 0.0851, 0.0956],
+    [0.6845, 0.6735, 0.6748],
+    [7.8790, 8.0568, 8.0540],
+    [0.0667, 0.0678, 0.0613],
+    [0.3354, 0.3023, 0.3058],
+    [0.3767, 0.3575, 0.3388],
+    [1.2356, 0.9725, 1.0842],
+    [1.0723, 1.0657, 1.0676],
+    [1.0918, 1.1146, 1.1810],
+    [0.2625, 0.2573, 0.2693],
+    [0.0767, 0.0745, 0.0825],
+    [0.0864, 0.0874, 0.0780],
+    [0.0515, 0.0521, 0.0534],
+    [0.1459, 0.1314, 0.1357],
+    [0.0520, 0.0382, 0.0409],
+    [0.0413, 0.0402, 0.0383],
+    [0.1223, 0.1212, 0.1203]
+  ]
+}
diff --git a/arc/run.sh b/arc/run.sh
new file mode 100755
index 000000000..8b8ac56ff
--- /dev/null
+++ b/arc/run.sh
@@ -0,0 +1,95 @@
+#!/bin/bash
+# Arc ClickBench Benchmark Runner
+# Queries Arc via HTTP API using Apache Arrow columnar format
+
+TRIES=3
+DATABASE="${DATABASE:-clickbench}"
+TABLE="${TABLE:-hits}"
+ARC_URL="${ARC_URL:-http://localhost:8000}"
+ARC_API_KEY="${ARC_API_KEY:-benchmark-test-key}"
+
+# Check if Arc is running
+echo "Checking if Arc is running at $ARC_URL..." >&2
+if ! curl -s -f "$ARC_URL/health" > /dev/null 2>&1; then
+    echo "Error: Arc is not running at $ARC_URL" >&2
+    echo "Please start Arc first or set ARC_URL environment variable" >&2
+    exit 1
+fi
+
+echo "Arc is running. Querying table: $DATABASE.$TABLE (Apache Arrow)" >&2
+echo "Using API key: ${ARC_API_KEY:0:20}..." >&2
+
+python3 << EOF
+import requests
+import time
+import sys
+
+try:
+    import pyarrow as pa
+except ImportError:
+    print("Error: pyarrow is required for Arrow format", file=sys.stderr)
+    print("Install with: pip install pyarrow", file=sys.stderr)
+    sys.exit(1)
+
+ARC_URL = "$ARC_URL"
+API_KEY = "$ARC_API_KEY"
+DATABASE = "$DATABASE"
+TABLE = "$TABLE"
+
+# Headers for API requests
+headers = {
+    "x-api-key": API_KEY,
+    "Content-Type": "application/json"
+}
+
+# Read queries
+with open('queries.sql') as f:
+    content = f.read()
+
+# Remove comment lines
+lines = [line for line in content.split('\n') if not line.strip().startswith('--')]
+clean_content = '\n'.join(lines)
+
+# Split by semicolons and filter empties
+queries = []
+for query in clean_content.split(';'):
+    query = query.strip()
+    if query:
+        queries.append(query)
+
+print(f"Running {len(queries)} queries via Apache Arrow API...", file=sys.stderr)
+
+# Run each query 3 times
+for i, query_sql in enumerate(queries, 1):
+    for run in range(3):
+        try:
+            start = time.perf_counter()
+
+            response = requests.post(
+                f"{ARC_URL}/query/arrow",
+                headers=headers,
+                json={"sql": query_sql},
+                timeout=300
+            )
+
+            if response.status_code == 200:
+                # Parse Arrow IPC stream to ensure data is received
+                reader = pa.ipc.open_stream(response.content)
+                arrow_table = reader.read_all()
+                elapsed = time.perf_counter() - start
+                print(f"{elapsed:.4f}")
+            else:
+                print("null")
+                if run == 0:
+                    print(f"Query {i} failed: {response.status_code} - {response.text[:200]}", file=sys.stderr)
+        except requests.exceptions.Timeout:
+            print("null")
+            if run == 0:
+                print(f"Query {i} timed out", file=sys.stderr)
+        except Exception as e:
+            print("null")
+            if run == 0:
+                print(f"Query {i} error: {e}", file=sys.stderr)
+
+print("Benchmark complete!", file=sys.stderr)
+EOF
diff --git a/arc/template.json b/arc/template.json
new file mode 100644
index 000000000..526bafb74
--- /dev/null
+++ b/arc/template.json
@@ -0,0 +1,6 @@
+{
+  "system": "Arc",
+  "proprietary": "no",
+  "tuned": "no",
+  "tags": ["Python", "time-series", "DuckDB", "Parquet", "columnar"]
+}