HCI Configuration Best Practices
Read the full document on Sangfor Cloud Platform (SCP)
The Sangfor HCI Configuration Best Practices Guide provides a holistic framework for IT professionals to optimize their Hyper-Converged Infrastructure (HCI) deployments. This essential document covers every critical layer of the infrastructure, ensuring maximum performance, reliability, and seamless scalability.
Core Infrastructure Foundations
Hardware: Emphasizes using identical physical hosts within a cluster to prevent instability, with clear prerequisites for CPU virtualization support, memory capacity, and storage redundancy (RAID).
BIOS Configuration: Details essential settings for CPU virtualization, Hyper-Threading, NUMA, and power management to ensure consistent, high-level performance.
Platform Optimization Strategies
Resource Management: Covers strategies to prevent resource starvation, including prohibiting memory overcommitment, reserving host resources for failover, and configuring VM cloning and anti-affinity policies.
Reliability & High Availability: Addresses HA, Dynamic Resource Extension (DRX), Distributed Resource Scheduler (DRS), data balancing, intelligent rate restriction, and data rebuilding protocols.
Network Architecture: Provides configuration guidelines for VXLAN, Management, and Edge networks, focusing on traffic isolation, link aggregation, and switch protocol settings.
aSAN (Advanced Storage Area Network): Outlines requirements for the storage network, cache SSDs, data disks, and optimal storage volume structuring.
Operational Management and Protection
Backup & Disaster Recovery: Details global backup policies, external backup storage usage, scheduling strategies, and snapshot-based backup for large VMs.
Alerting & Management: Recommends enabling comprehensive alerts across all components (Host, Network, Storage, VMs, Containers, Security, etc.) and provides guidance on version control, upgrade strategies, patch management, and routine health checks using the aCheck tool.
Virtual Machine Parameter Tuning
The guide concludes with in-depth explanations for fine-tuning VM parameters across different business scenarios (databases, web servers, Kubernetes clusters, etc.), covering CPU tuning (NUMA binding, Host CPU pass-through, CPU exclusive mode, clock speed limits, Hot Add), memory tuning (reclaiming, Huge Pages, Hot Add, allocation modes), and storage and network I/O (Virtio Disk, disk I/O limits, NIC traffic limits, adapter model, Jumbo Frame, VMtool, HA, and High Performance mode).
By adhering to these guidelines across all modules, organizations can build a highly resilient, secure, and performant Sangfor HCI infrastructure tailored to their specific enterprise needs.