Performance Tuning

Learn how to optimize Edera’s performance through memory management, storage optimization, device configuration, and advanced features for maximum efficiency in your containerized workloads.

Memory Optimization

Dynamic Memory Ballooning

Edera’s dynamic memory ballooning automatically adjusts memory allocation between configured minimum and maximum values based on actual process usage, providing optimal resource utilization.

How Dynamic Ballooning Works

• Automatic Scaling: Memory resizes based on usage patterns
• Downscaling: If ≥90% memory unused, zone resizes to half current allocation
• Upscaling: If ≥70% memory used, increases to current max or current usage + 80 MB
• Check Intervals: Start at 100ms, then stabilize to 2-second intervals

Default Memory Settings

# Default configuration
max_memory: 1024M
target_memory: 300M
min_memory: 300M (minimum allowed)

Configuration Options

Via Pod Annotations:

annotations:
    dev.edera/initial-memory-request: "4096"
    dev.edera/resource-policy: "dynamic"
spec:
  runtimeClassName: edera

Via CLI:

# Update memory settings
protect zone update-resources --target-memory 4096 my-zone

Memory Best Practices

• Use Dynamic Mode: Recommended for most workloads (default)
• Set Appropriate Limits: Configure max memory based on workload requirements
• Monitor Usage: Use Edera metrics to track memory efficiency
• Avoid Static Mode: Only use for specific advanced use cases

Memory Ballooning Guide

Storage Optimization

Scratch Disks

Scratch disks provide temporary, high-speed storage volumes that decouple disk storage from memory requirements, enabling high-disk, low-memory workloads with improved performance.

Scratch Disk Benefits

• Performance: Better I/O performance than memory-backed storage
• Flexibility: Support for high-disk, low-memory workloads
• Resource Efficiency: Separate disk and memory allocation
• Scalability: Flexible zone sizing options

Storage Best Practices

• Use Block Devices: For highest I/O performance
• Size Appropriately: Match scratch disk size to workload requirements
• Remember Non-Persistence: All data is wiped on zone restart
• Monitor I/O: Track disk usage patterns with Edera metrics

Scratch Disks Setup

Device Optimization

Direct Device Passthrough

Claim and pass through specific hardware devices directly to zones for maximum performance and specialized hardware access.

Supported Device Types

• Block Devices: NVMe drives, SSDs for high-performance storage
• PCI Hardware: Direct hardware access for specialized workloads
• SR-IOV Virtual Functions: Network interface virtualization (v1.3.1+)

When this is useful:

• You want to use a specific block device (for example, NVMe drive)
• You’re working with passthrough hardware like PCI-attached hardware
• You’re experimenting with SR-IOV and need zone-level access to virtual functions (VFs)

Device Claiming

Advanced Performance Features

eBPF Integration

Edera supports eBPF by default for advanced performance monitoring and optimization without additional tooling or kernel patches.

As long as you’re using an Edera zone kernel that supports eBPF (via the dev.edera/kernel pod annotation), you’re good to go.

For more information follow our eBPF integration guide.

Additional Resources

For detailed implementation guides:

• Advanced Guides

Need help optimizing your specific workload? Contact support@edera.dev for performance analysis and tuning assistance.