jester/knowledge-base
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
88c0258a9e
knowledge-base/ZeroLagHub_Infrastructure_Specifications.md
jester 5a5ad001af Upload files to "/"
2025-12-13 16:40:48 +00:00

14 KiB
Raw Blame History

🏗️ ZeroLagHub Infrastructure - Complete Specifications

Last Updated: December 7, 2025
Provider: GTHost
Cost: $109/month
Status: Production Infrastructure

🖥️ Dedicated Server Hardware

Server Platform

Model: Supermicro 2029TP-HC1R (hot-swap HDD chassis)
Form Factor: Enterprise rackmount server

CPU

Model: Intel Xeon Silver 4116
Cores: 12 cores / 24 threads
Clock Speed: 2.1 GHz base, 3.0 GHz turbo
Architecture: Intel Skylake-SP (Server Processor)
TDP: 85W
Cache: 16.5 MB L3

Performance:

  • Single-threaded: Excellent for game servers
  • Multi-threaded: Handles 11 VMs + 75-100 containers
  • Turbo boost ensures responsive provisioning

Memory (RAM)

Configuration: 6 x 32GB DIMMs
Total: 192 GB
Type: Hynix DDR4 RDIMM (Registered)
Speed: 2400 MHz
ECC: Yes (Error-Correcting Code)

Benefits:

  • ECC protects against memory corruption
  • Registered DIMMs = enterprise reliability
  • 192GB = ample headroom for VM overhead + game servers

Storage

Configuration: 2 x 1.92TB SSD
Model: Samsung PM863 (Enterprise SSD)
Total Capacity: 3.84 TB raw
Available: ~1.8 TB (after Proxmox + VMs + overhead)
Interface: SATA (likely in RAID configuration)

Samsung PM863 Specs:

  • Enterprise-grade datacenter SSD
  • Optimized for mixed workload
  • Power Loss Protection (PLP)
  • High endurance rating

RAID Configuration (Likely):

  • RAID 1 (mirrored) for redundancy
  • Or ZFS RAID-Z for Proxmox storage

Network

Bandwidth: 300 Mbit/s (37.5 MB/s)
Metering: Unmetered (no bandwidth caps)
Uplink: Enterprise datacenter connection

Performance Assessment:

  • 300 Mbit/s = 30-40 concurrent Minecraft players comfortably
  • Unmetered = no surprise overage charges
  • Sufficient for soft launch, may need upgrade at scale

Operating System

OS: Proxmox VE 8 (64-bit)
Base: Debian 12 "Bookworm"
Kernel: Linux 6.2+

📊 Partition Layout

Disk Partitions

/boot:  1024 MB (1 GB)  - Boot partition
/swap:  2048 MB (2 GB)  - Swap space
/root:  Auto size       - Main Proxmox system + VM storage
                          (~3.8 TB usable after RAID)

Storage Allocation (Estimated):

Total:          3.84 TB raw (2 x 1.92TB)
RAID overhead:  -0.04 TB (metadata, alignment)
                --------
Available:      3.80 TB

Proxmox OS:     -0.10 TB (Proxmox + system)
VM Disks:       -1.80 TB (11 VMs, templates)
LXC Containers: -0.10 TB (current containers)
                --------
Free Space:     1.80 TB (for game servers + growth)

🎯 Performance Characteristics

CPU Capabilities

Per-Core Performance: ⭐⭐⭐⭐ (Excellent for game servers)

  • Minecraft servers are single-threaded
  • 3.0 GHz turbo provides snappy performance
  • 12 cores = 12 simultaneous high-performance game servers

Multi-Core Throughput: ⭐⭐⭐⭐⭐ (Excellent for hosting)

  • 24 threads handle VM overhead efficiently
  • Can run 11 Proxmox VMs + 75-100 LXC containers
  • Provisioning operations don't impact running servers

Virtualization: ⭐⭐⭐⭐⭐ (Native support)

  • Intel VT-x + VT-d enabled
  • Hardware-accelerated virtualization
  • LXC containers = near-native performance

Memory Capabilities

Capacity: 192 GB = Excellent for scale

  • Average Minecraft server: 2-4 GB
  • 192 GB / 4 GB = 48 simultaneous 4GB servers
  • Or 96 lightweight 2GB servers

Speed: 2400 MHz DDR4 = Good (not bleeding edge, but sufficient)

  • DDR4-2400 provides adequate bandwidth for game hosting
  • ECC ensures data integrity under load

Reliability: ECC RDIMM = Enterprise-grade

  • Detects and corrects memory errors
  • Critical for 24/7 uptime

Storage Capabilities

Capacity: 3.84 TB = Very Good for initial scale

  • Each Minecraft server: 1-5 GB (world size varies)
  • Can host 300-500 Minecraft servers comfortably
  • 1.8 TB free = room for significant growth

Performance: Samsung PM863 = Excellent for workload

  • Random IOPS: ~10,000 read, ~2,000 write
  • Sequential: 520 MB/s read, 485 MB/s write
  • Perfect for database + game world I/O

Reliability: Enterprise SSD = Excellent

  • Power Loss Protection prevents corruption
  • Rated for 1.3 PB writes (years of 24/7 operation)
  • RAID 1 (likely) provides redundancy

Network Capabilities

Bandwidth: 300 Mbit/s = Adequate for soft launch

  • Minecraft player: ~0.5-1 Mbit/s
  • 300 Mbit/s = 30-60 players (conservative estimate)
  • Unmetered = no bandwidth overage charges

Upgrade Path: GTHost likely offers 1 Gbps upgrades

  • 1 Gbps would support 100-200 players
  • Consider upgrade when approaching 40+ concurrent players

🏗️ Current Resource Allocation

VM Resource Breakdown (11 VMs)

Hypervisor Overhead:  ~8 GB RAM, 2 CPU cores

Critical Production:
├─ VM 100 (zlh-panel):       4 GB RAM, 2 cores, 32 GB disk
├─ VM 103 (zlh-api):         4 GB RAM, 2 cores, 32 GB disk
└─ VM 101 (zlh-wings):       8 GB RAM, 4 cores, 64 GB disk

Platform Services:
├─ VM 102 (zlh-portal):      4 GB RAM, 2 cores, 32 GB disk
├─ VM 104 (zlh-monitor):     8 GB RAM, 2 cores, 64 GB disk
└─ VM 1002 (zlh-proxy):      2 GB RAM, 1 core, 16 GB disk

Network Layer:
├─ VM 1000 (zlh-router):     4 GB RAM, 2 cores, 32 GB disk
├─ VM 1006 (zpack-router):   4 GB RAM, 2 cores, 32 GB disk
└─ VM 1001 (zlh-dns):        2 GB RAM, 1 core, 16 GB disk

Development/Support:
├─ VM 300 (zlh-panel-dev):   4 GB RAM, 2 cores, 32 GB disk
└─ VM 2000 (zlh-ci):         4 GB RAM, 2 cores, 32 GB disk

Backup:
└─ VM [zlh-back]:            8 GB RAM, 2 cores, 128 GB disk

TOTAL VM ALLOCATION:         56 GB RAM, 24 cores, ~512 GB disk

Available for Game Servers

RAM Available:    192 GB - 56 GB (VMs) - 8 GB (overhead) = 128 GB
CPU Available:    24 threads - 24 (VM allocation) = 0 (shared)
Disk Available:   1.8 TB free

Game Server Capacity (Conservative):
├─ 2GB servers:   64 simultaneous servers
├─ 4GB servers:   32 simultaneous servers
└─ 8GB servers:   16 simultaneous servers

Developer Environment Capacity:
├─ 2GB dev envs:  64 simultaneous environments
└─ 4GB dev envs:  32 simultaneous environments

Note: CPU is oversubscribed (common in hosting) since most game servers idle at <20% CPU usage. Turbo boost ensures good single-thread performance when needed.

📈 Capacity & Scaling Projections

Current Capacity (As Deployed)

Game Servers: 30-50 active servers with current VM allocation
Developer Environments: 75-100 environments (documented capacity)
Concurrent Players: 30-60 players (network limited)

Optimized Capacity (With Tuning)

Game Servers: 60-80 active servers (after VM consolidation)
Developer Environments: 100-150 environments
Concurrent Players: Still 30-60 (network bottleneck)

Maximum Theoretical Capacity

Game Servers: 128 lightweight servers (if only game hosting, no dev)
Developer Environments: 192 environments (if only dev, no games)
Storage: 300-500 servers before storage exhaustion

Limiting Factors:

  1. Network (300 Mbit/s) - limits concurrent players
  2. RAM (192 GB) - limits concurrent heavy servers
  3. Storage (1.8 TB free) - limits total servers

💰 Cost Analysis

Current Infrastructure Cost

Monthly: $109 GTHost dedicated server
Annually: $1,308

Cost per Resource:

  • Per GB RAM: $0.57/month ($109 ÷ 192 GB)
  • Per CPU core: $9.08/month ($109 ÷ 12 cores)
  • Per TB storage: $28.39/month ($109 ÷ 3.84 TB)

Competitive Analysis

AWS Equivalent (m5.2xlarge + storage + bandwidth):

  • 8 vCPU, 32 GB RAM, 1 TB storage, 1 Gbps
  • Cost: ~$300-400/month

Hetzner Dedicated (Similar specs):

  • 12 core Xeon, 128 GB RAM, 2x2TB SSD
  • Cost: ~$100/month (but higher network costs)

GTHost Value: ⭐⭐⭐⭐⭐ Excellent

  • 40-60% cheaper than AWS
  • Competitive with Hetzner
  • Unmetered bandwidth (key advantage)

🎯 Competitive Advantages

1. LXC Performance

  • Host hardware enables 20-30% better performance vs Docker
  • Intel Xeon Silver 4116 single-thread performance excellent for games

2. Resource Density

  • 192 GB RAM supports 30-50 simultaneous 4GB servers
  • Competitors typically offer 64-128 GB at this price point

3. Storage Performance

  • Samsung PM863 enterprise SSDs outperform consumer SSDs
  • Power Loss Protection prevents world corruption
  • Hot-swap chassis enables maintenance without downtime

4. Network

  • Unmetered = no bandwidth surprises
  • 300 Mbit/s adequate for soft launch
  • Upgrade path available when needed

⚠️ Identified Constraints

1. Network Bandwidth (Current Bottleneck)

  • 300 Mbit/s limits to 30-60 concurrent players
  • Recommendation: Monitor bandwidth usage, upgrade to 1 Gbps when approaching 40 players
  • Upgrade Cost: Likely +$20-50/month for 1 Gbps

2. CPU Oversubscription

  • 24 threads allocated to VMs, but most VMs idle
  • Game servers share CPU via time-slicing
  • Risk: If all servers spike simultaneously, performance degrades
  • Mitigation: Limit concurrent servers to 40-50 until load testing proves higher safe

3. Storage Growth

  • 1.8 TB free supports 300-500 servers
  • Each server grows over time (world expansion)
  • Recommendation: Monitor disk usage, plan expansion at 70% utilization
  • Expansion Options: Add external storage or upgrade to larger SSDs

🔧 Optimization Opportunities

Immediate Optimizations (No Cost)

  1. VM Consolidation

    • Merge zlh-panel-dev into zlh-panel (save 4 GB RAM, 2 cores)
    • Merge zlh-proxy into zlh-router (save 2 GB RAM, 1 core)
    • Gain: 6 GB RAM, 3 cores for game servers

  2. LXC Over VMs

    • Convert lightweight VMs to LXC containers
    • Example: zlh-dns, zlh-proxy candidates
    • Gain: Lower overhead, faster provisioning

  3. Memory Ballooning

    • Enable KSM (Kernel Same-page Merging) on Proxmox
    • Deduplicate identical memory pages
    • Gain: 5-10% more available RAM

Paid Optimizations (Consider at Scale)

  1. Network Upgrade: 1 Gbps uplink (+$20-50/month)

    • Removes player concurrency bottleneck
    • Enables 100-200 player capacity

  2. Storage Expansion: Add 4TB NVMe (+$50/month)

    • Doubles storage to ~6 TB total
    • Supports 600-1000 servers

  3. Cloudflare Enterprise (+$200/month)

    • DDoS protection for game traffic
    • CDN for static assets
    • Worth it at 100+ servers

📊 Hardware Lifecycle

Current Status (December 2025)

Server Age: Unknown (likely 1-3 years based on Xeon Silver 4116 era)
Expected Lifespan: 5-7 years for enterprise server
Remaining Life: Likely 3-5 years

Components:

  • CPU: Xeon Silver 4116 (2017 release) - still very capable
  • RAM: DDR4-2400 (current gen, plenty of life)
  • SSD: Samsung PM863 (enterprise grade, high endurance)

Upgrade Path (Future)

Year 1-2 (Current plan):

  • Optimize existing hardware
  • Minor network upgrades if needed

Year 3-4 (Growth phase):

  • Consider second dedicated server
  • Load balance across servers
  • Geographic distribution

Year 5+ (Scale phase):

  • Migrate to colocation or cloud
  • Multi-datacenter deployment

🛡️ Reliability Features

Hardware Reliability

✅ ECC Memory - Corrects single-bit errors automatically
✅ Enterprise SSDs - Power Loss Protection, high endurance
✅ Hot-Swap Chassis - Replace drives without shutdown
✅ Redundant Power (likely) - Supermicro chassis typically dual PSU

Software Reliability

✅ Proxmox High Availability - VM failover (if configured)
✅ PBS Backup - Incremental backups to Backblaze B2
✅ LXC Snapshots - Fast rollback capability
✅ RAID Mirroring (likely) - Disk failure protection

Network Reliability

✅ Datacenter Uptime - GTHost likely 99.9%+ SLA
✅ Unmetered Bandwidth - No throttling during spikes
⚠️ Single Uplink - No network redundancy (acceptable for price point)

🎯 Summary & Recommendations

Hardware Assessment: ⭐⭐⭐⭐ Very Good for Use Case

Strengths:

  • Excellent CPU for game server hosting (Xeon Silver 4116)
  • Abundant RAM (192 GB = 30-50 servers)
  • Enterprise storage (Samsung PM863 + hot-swap)
  • Unmetered bandwidth (no surprise charges)
  • Great value ($109/month for these specs)

Limitations:

  • Network bandwidth (300 Mbit/s = 30-60 players)
  • Storage growth constraint (monitor usage)
  • CPU oversubscription (limit concurrent servers initially)

Recommendations

Now (Launch Phase):

  1. ✅ Deploy on current hardware - adequate for soft launch
  2. ✅ Limit to 40-50 concurrent servers initially
  3. ✅ Monitor bandwidth, RAM, and disk usage

Month 1-3 (Early Growth):

  1. 🔧 Optimize VM allocation (consolidate where possible)
  2. 🔧 Implement aggressive monitoring
  3. 🔧 Consider 1 Gbps network upgrade if approaching 40 players

Month 6-12 (Scale Phase):

  1. 📈 Evaluate storage expansion based on usage
  2. 📈 Consider second server for geographic distribution
  3. 📈 Implement Cloudflare Enterprise for DDoS protection

Capacity Targets by Phase

Soft Launch (Month 1-3): 20-30 servers, 10-20 players
Public Launch (Month 3-6): 40-50 servers, 30-40 players
Growth Phase (Month 6-12): 60-80 servers, 60-100 players (with 1 Gbps upgrade)
Scale Phase (Month 12+): 100+ servers, multi-server deployment

✅ Conclusion

Status: Infrastructure is production-ready for ZeroLagHub launch.

Key Points:

  • Hardware specifications are excellent for initial scale
  • 192 GB RAM supports 30-50 game servers
  • Storage capacity adequate for 300-500 servers
  • Network bandwidth is current bottleneck (acceptable for soft launch)
  • Cost-effective ($109/month for enterprise-grade hardware)

Green Light: ✅ Launch when platform development complete (currently 85% ready).

Last Updated: December 7, 2025
Source: GTHost server specifications + ZeroLagHub infrastructure analysis