# Time-Series Database Schema Specification

## Overview

This document specifies the time-series database schema using ClickHouse or TimescaleDB for storing mempool data, metrics, and analytics time-series data.

## Technology Choice

**Option 1: TimescaleDB** (PostgreSQL extension)
- Pros: PostgreSQL compatibility, SQL interface, easier integration
- Cons: Less optimized for very high throughput

**Option 2: ClickHouse**
- Pros: Very high performance, columnar storage, excellent compression
- Cons: Different SQL dialect, separate infrastructure

**Recommendation**: Start with TimescaleDB for easier integration, migrate to ClickHouse if needed for scale.

## TimescaleDB Schema

### Mempool Transactions Table

**Table**: `mempool_transactions`

```sql
CREATE TABLE mempool_transactions (
    time TIMESTAMPTZ NOT NULL,
    chain_id INTEGER NOT NULL,
    hash VARCHAR(66) NOT NULL,
    from_address VARCHAR(42) NOT NULL,
    to_address VARCHAR(42),
    value NUMERIC(78, 0),
    gas_price BIGINT,
    max_fee_per_gas BIGINT,
    max_priority_fee_per_gas BIGINT,
    gas_limit BIGINT,
    nonce BIGINT,
    input_data_length INTEGER,
    first_seen TIMESTAMPTZ NOT NULL,
    status VARCHAR(20) DEFAULT 'pending', -- 'pending', 'confirmed', 'dropped'
    confirmed_block_number BIGINT,
    confirmed_at TIMESTAMPTZ,
    PRIMARY KEY (time, chain_id, hash)
);

SELECT create_hypertable('mempool_transactions', 'time');

CREATE INDEX idx_mempool_chain_hash ON mempool_transactions(chain_id, hash);
CREATE INDEX idx_mempool_chain_from ON mempool_transactions(chain_id, from_address);
CREATE INDEX idx_mempool_chain_status ON mempool_transactions(chain_id, status, time);
```

**Retention Policy**: 7 days for detailed data, aggregates for longer periods

### Network Metrics Table

**Table**: `network_metrics`

```sql
CREATE TABLE network_metrics (
    time TIMESTAMPTZ NOT NULL,
    chain_id INTEGER NOT NULL,
    block_number BIGINT,
    tps DOUBLE PRECISION, -- Transactions per second
    gps DOUBLE PRECISION, -- Gas per second
    avg_gas_price BIGINT,
    pending_transactions INTEGER,
    block_time_seconds DOUBLE PRECISION,
    PRIMARY KEY (time, chain_id)
);

SELECT create_hypertable('network_metrics', 'time');

CREATE INDEX idx_network_metrics_chain_time ON network_metrics(chain_id, time DESC);
```

**Aggregation**: Pre-aggregate to 1-minute, 5-minute, 1-hour intervals

### Gas Price History Table

**Table**: `gas_price_history`

```sql
CREATE TABLE gas_price_history (
    time TIMESTAMPTZ NOT NULL,
    chain_id INTEGER NOT NULL,
    block_number BIGINT,
    min_gas_price BIGINT,
    max_gas_price BIGINT,
    avg_gas_price BIGINT,
    p25_gas_price BIGINT, -- 25th percentile
    p50_gas_price BIGINT, -- 50th percentile (median)
    p75_gas_price BIGINT, -- 75th percentile
    p95_gas_price BIGINT, -- 95th percentile
    p99_gas_price BIGINT, -- 99th percentile
    PRIMARY KEY (time, chain_id)
);

SELECT create_hypertable('gas_price_history', 'time');
```

### Address Activity Metrics Table

**Table**: `address_activity_metrics`

```sql
CREATE TABLE address_activity_metrics (
    time TIMESTAMPTZ NOT NULL,
    chain_id INTEGER NOT NULL,
    address VARCHAR(42) NOT NULL,
    transaction_count INTEGER,
    received_count INTEGER,
    sent_count INTEGER,
    total_received NUMERIC(78, 0),
    total_sent NUMERIC(78, 0),
    PRIMARY KEY (time, chain_id, address)
);

SELECT create_hypertable('address_activity_metrics', 'time', 
    chunk_time_interval => INTERVAL '1 day');

CREATE INDEX idx_address_activity_chain_address ON address_activity_metrics(chain_id, address, time DESC);
```

**Aggregation**: Pre-aggregate to hourly/daily for addresses

## ClickHouse Schema (Alternative)

### Mempool Transactions Table

```sql
CREATE TABLE mempool_transactions (
    time DateTime('UTC') NOT NULL,
    chain_id UInt32 NOT NULL,
    hash String NOT NULL,
    from_address String NOT NULL,
    to_address Nullable(String),
    value Decimal128(0),
    gas_price UInt64,
    max_fee_per_gas Nullable(UInt64),
    max_priority_fee_per_gas Nullable(UInt64),
    gas_limit UInt64,
    nonce UInt64,
    input_data_length UInt32,
    first_seen DateTime('UTC') NOT NULL,
    status String DEFAULT 'pending',
    confirmed_block_number Nullable(UInt64),
    confirmed_at Nullable(DateTime('UTC'))
) ENGINE = MergeTree()
PARTITION BY toYYYYMM(time)
ORDER BY (chain_id, time, hash)
TTL time + INTERVAL 7 DAY; -- Auto-delete after 7 days
```

## Data Retention and Aggregation

### Retention Policies

**Raw Data**:
- Mempool transactions: 7 days
- Network metrics: 30 days
- Gas price history: 90 days
- Address activity: 30 days

**Aggregated Data**:
- 1-minute aggregates: 90 days
- 5-minute aggregates: 1 year
- 1-hour aggregates: 5 years
- Daily aggregates: Indefinite

### Continuous Aggregates (TimescaleDB)

```sql
-- 1-minute network metrics aggregate
CREATE MATERIALIZED VIEW network_metrics_1m
WITH (timescaledb.continuous) AS
SELECT
    time_bucket('1 minute', time) AS bucket,
    chain_id,
    AVG(tps) AS avg_tps,
    AVG(gps) AS avg_gps,
    AVG(avg_gas_price) AS avg_gas_price,
    AVG(pending_transactions) AS avg_pending_tx
FROM network_metrics
GROUP BY bucket, chain_id;

-- Add refresh policy
SELECT add_continuous_aggregate_policy('network_metrics_1m',
    start_offset => INTERVAL '1 hour',
    end_offset => INTERVAL '1 minute',
    schedule_interval => INTERVAL '1 minute');
```

## Query Patterns

### Recent Mempool Transactions

```sql
SELECT * FROM mempool_transactions
WHERE chain_id = 138
  AND time > NOW() - INTERVAL '1 hour'
  AND status = 'pending'
ORDER BY time DESC
LIMIT 100;
```

### Gas Price Statistics

```sql
SELECT 
    time_bucket('5 minutes', time) AS bucket,
    AVG(avg_gas_price) AS avg_gas_price,
    PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY avg_gas_price) AS median_gas_price
FROM gas_price_history
WHERE chain_id = 138
  AND time > NOW() - INTERVAL '24 hours'
GROUP BY bucket
ORDER BY bucket DESC;
```

### Network Throughput

```sql
SELECT 
    time_bucket('1 minute', time) AS bucket,
    AVG(tps) AS avg_tps,
    MAX(tps) AS max_tps
FROM network_metrics
WHERE chain_id = 138
  AND time > NOW() - INTERVAL '1 hour'
GROUP BY bucket
ORDER BY bucket DESC;
```

## References

- Mempool Service: See `../mempool/mempool-service.md`
- Observability: See `../observability/metrics-monitoring.md`