Cortex Logo

Cortex

Semantics

Cache

Control how metric results are stored and retrieved to improve performance and reduce database load.

Overview

Caching stores the results of metric queries to avoid re-executing expensive database operations. This is especially important for complex metrics that process large amounts of data or perform heavy calculations.

Diagram showing how caching works in the Cortex system

Use Cases

Caching is used for:

  • Performance Optimization: Speed up frequently accessed metrics
  • Cost Reduction: Reduce database load and query costs
  • User Experience: Provide faster response times for dashboards
  • Resource Management: Reduce server load during peak usage
  • Data Freshness: Balance performance with data accuracy

Specification

Type: CachePreference

Result cache configuration for metrics and requests.

{
    "cache": {
        "enabled": true,
        "ttl": 3600
    }
}

Configuration

enabled

Type: bool
Default: True
Required: No

Whether result caching is active. When enabled, query results will be cached for the specified TTL.

{
    "enabled": true
}

ttl

Type: Optional[int]
Default: None
Required: No

Time-to-live in seconds. How long cached results should be stored before expiring.

{
    "ttl": 3600
}

Strategies

Short-term Caching (High Frequency, Low Latency)

{
    "cache": {
        "enabled": true,
        "ttl": 300  # 5 minutes
    }
}

Best for: Real-time dashboards, frequently updated metrics

Medium-term Caching (Balanced Performance)

{
    "cache": {
        "enabled": true,
        "ttl": 3600  # 1 hour
    }
}

Best for: Standard business metrics, regular reports

Long-term Caching (Historical Data)

{
    "cache": {
        "enabled": true,
        "ttl": 86400  # 24 hours
    }
}

Best for: Historical analysis, archived data

No Caching (Real-time Data)

{
    "cache": {
        "enabled": false
    }
}

Best for: Critical real-time metrics, live monitoring

Key Generation

Automatic Key Generation

Cache keys are automatically generated based on:

  • Metric ID
  • Parameter values
  • Query filters
  • User context
  • Data source version

Custom Cache Keys

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "key_template": "metric_{metric_id}_{user_segment}_{date_range}"
    }
}

Invalidation

Time-based Invalidation

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "invalidation_strategy": "time_based"
    }
}

Data-based Invalidation

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "invalidation_strategy": "data_based",
        "invalidation_conditions": [
            "orders.created_at > last_cache_time",
            "users.updated_at > last_cache_time"
        ]
    }
}

Manual Invalidation

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "invalidation_strategy": "manual"
    }
}

Storage Options

Memory Cache

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "storage": "memory",
        "max_size": "1GB"
    }
}

Redis Cache

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "storage": "redis",
        "redis_url": "redis://localhost:6379",
        "redis_db": 0
    }
}

Database Cache

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "storage": "database",
        "cache_table": "metric_cache"
    }
}

Cache Performance Tuning

Memory Management

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "max_memory": "2GB",
        "eviction_policy": "lru"
    }
}

Compression

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "compression": true,
        "compression_level": 6
    }
}

Cache Warming

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "warmup": true,
        "warmup_schedule": "0 0 * * *"  # Daily at midnight
    }
}

Cache Monitoring

Cache Hit Rate

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "monitoring": {
            "track_hit_rate": true,
            "alert_threshold": 0.8
        }
    }
}

Cache Size Monitoring

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "monitoring": {
            "track_size": true,
            "max_size_alert": "1GB"
        }
    }
}

Performance Metrics

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "monitoring": {
            "track_performance": true,
            "slow_query_threshold": "100ms"
        }
    }
}

Common Cache Patterns

Dashboard Caching

{
    "cache": {
        "enabled": true,
        "ttl": 300,  # 5 minutes
        "description": "Dashboard metrics refresh every 5 minutes"
    }
}

Report Caching

{
    "cache": {
        "enabled": true,
        "ttl": 3600,  # 1 hour
        "description": "Reports are cached for 1 hour"
    }
}

Historical Data Caching

{
    "cache": {
        "enabled": true,
        "ttl": 86400,  # 24 hours
        "description": "Historical data doesn't change often"
    }
}

Real-time Metrics

{
    "cache": {
        "enabled": false,
        "description": "Real-time metrics should not be cached"
    }
}

Best Practices

Choose Appropriate TTL

  • High-frequency metrics: 5-15 minutes
  • Standard metrics: 1-4 hours
  • Historical data: 24 hours or more
  • Real-time data: No caching

Consider Data Freshness

{
    "cache": {
        "enabled": true,
        "ttl": 1800,  # 30 minutes
        "description": "Balance between performance and data freshness"
    }
}

Monitor Performance

  • Track hit rates
  • Monitor cache size
  • Alert on performance issues
  • Regular cache cleanup

Use Cache Warming

{
    "cache": {
        "enabled": true,
        "ttl": 3600,
        "warmup": true,
        "warmup_queries": [
            "popular_metrics",
            "dashboard_queries"
        ]
    }
}

Troubleshooting

Low Hit Rate

  • Check TTL settings
  • Verify cache key generation
  • Review query patterns
  • Consider cache warming

High Memory Usage

  • Reduce TTL
  • Implement compression
  • Use LRU eviction
  • Consider distributed caching

Stale Data

  • Reduce TTL
  • Implement data-based invalidation
  • Use cache versioning
  • Manual cache refresh

Video Tutorial

Video: How to configure caching for optimal performance in the Cortex dashboard

Performance Dashboard

Screenshot of the cache performance monitoring dashboard

Configuration Examples

E-commerce Metrics

{
    "cache": {
        "enabled": true,
        "ttl": 900,  # 15 minutes
        "description": "E-commerce metrics change frequently"
    }
}

Financial Reports

{
    "cache": {
        "enabled": true,
        "ttl": 14400,  # 4 hours
        "description": "Financial reports are generated periodically"
    }
}

User Analytics

{
    "cache": {
        "enabled": true,
        "ttl": 1800,  # 30 minutes
        "description": "User analytics balance performance and freshness"
    }
}

Related Topics

Parameters

Enable dynamic query generation by allowing metrics to accept runtime values, making metrics flexible and reusable.

Refresh

Control when pre-aggregated data is updated to ensure analytics stay current while optimizing performance.