Skip to content

README.md

Latest commit

 

History

History
114 lines (89 loc) · 3.7 KB

README.md

File metadata and controls

114 lines (89 loc) · 3.7 KB

Storage Infrastructure

This directory contains the storage configuration for the Kubernetes cluster, built on OpenEBS with Mayastor for high-performance container-attached storage.

Components

OpenEBS Configuration

  • openebs.yaml - OpenEBS operator and Mayastor engine deployment
  • disks.yaml - Disk pool configuration for storage nodes

Storage Classes

  • cache.yaml - Fast SSD storage class for caching and temporary data
  • crucial.yaml - High-performance NVMe storage class for databases
  • media.yaml - Large capacity storage class for media files

Architecture

Storage Nodes

The cluster includes two dedicated storage nodes (storage1, storage2) with:

  • Bonded Networking: Redundant network connections for high availability
  • Hugepages: Optimized memory allocation for Mayastor performance
  • Direct-attached Storage: NVMe SSDs for low-latency data access
  • NUMA Awareness: CPU and memory locality optimization

Mayastor Engine

Mayastor provides:

  • NVMe-over-TCP: High-performance storage protocol
  • Replication: Synchronous data replication across nodes
  • Thin Provisioning: Efficient storage space utilization
  • Snapshot Support: Point-in-time data recovery

Storage Classes

Cache Storage (mayastor-cache)

  • Use Case: Redis, temporary data, build caches
  • Performance: Ultra-low latency SSD storage
  • Replication: Single replica for maximum performance
  • Capacity: Optimized for small, frequent I/O operations

Crucial Storage (mayastor-crucial)

  • Use Case: Databases, critical application data
  • Performance: High IOPS NVMe storage
  • Replication: Multi-replica for data safety
  • Consistency: Synchronous replication across storage nodes

Media Storage (mayastor-media)

  • Use Case: Media files, backups, large datasets
  • Performance: Balanced throughput and capacity
  • Replication: Configurable based on data importance
  • Capacity: Large volume support for bulk storage

Usage

Requesting Storage

Applications can request storage using persistent volume claims:

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: database-storage
spec:
  accessModes:
    - ReadWriteOnce
  storageClassName: openebs-crucial
  resources:
    requests:
      storage: 10Gi

Storage Class Selection

Choose storage classes based on workload requirements:

  • Database workloads: openebs-crucial for consistency and performance
  • Cache layers: openebs-cache for maximum speed
  • Media storage: openebs-media for capacity and cost efficiency

Monitoring

Storage infrastructure is monitored through:

  • OpenEBS Metrics: Pool utilization, IOPS, latency
  • Node Metrics: Disk usage, network throughput
  • SigNoz Dashboards: Storage performance visualization
  • Alerting: Automatic notifications for storage issues

Backup and Recovery

Snapshots

Mayastor supports volume snapshots for:

  • Point-in-time recovery
  • Data migration between environments
  • Testing with production data copies

Replication

Data is automatically replicated across storage nodes to ensure:

  • High availability during node maintenance
  • Protection against hardware failures
  • Zero-downtime storage operations

Performance Tuning

Node Configuration

Storage nodes are optimized with:

  • Hugepages allocation for Mayastor
  • CPU isolation for storage workloads
  • Network bonding for redundancy
  • NUMA-aware resource allocation

I/O Optimization

  • NVMe drives configured for maximum performance
  • Queue depth optimization for workload patterns
  • Latency-sensitive workload prioritization
  • Background compaction scheduling