Game Servers

Game Servers

Server Architecture Patterns

┌─────────────────────────────────────────────────────────────┐
│                    SERVER ARCHITECTURES                      │
├─────────────────────────────────────────────────────────────┤
│  DEDICATED SERVER:                                           │
│  • Server runs game simulation                              │
│  • Clients send inputs, receive state                       │
│  • Best security and consistency                            │
│  • Higher infrastructure cost                               │
├─────────────────────────────────────────────────────────────┤
│  LISTEN SERVER:                                              │
│  • One player hosts the game                                │
│  • Free infrastructure                                      │
│  • Host has advantage (no latency)                          │
│  • Session ends if host leaves                              │
├─────────────────────────────────────────────────────────────┤
│  RELAY SERVER:                                               │
│  • Routes packets between peers                             │
│  • No game logic on server                                  │
│  • Good for P2P with NAT traversal                         │
│  • Less secure than dedicated                               │
└─────────────────────────────────────────────────────────────┘

Scalable Architecture

SCALABLE GAME BACKEND:
┌─────────────────────────────────────────────────────────────┐
│                     GLOBAL LOAD BALANCER                     │
│                            ↓                                 │
├─────────────────────────────────────────────────────────────┤
│                      GATEWAY SERVERS                         │
│            Authentication, Routing, Rate Limiting           │
│                            ↓                                 │
├─────────────────────────────────────────────────────────────┤
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐        │
│  │ MATCHMAKING │  │   LOBBY     │  │   SOCIAL    │        │
│  │   SERVICE   │  │   SERVICE   │  │   SERVICE   │        │
│  └─────────────┘  └─────────────┘  └─────────────┘        │
│                            ↓                                 │
├─────────────────────────────────────────────────────────────┤
│              GAME SERVER ORCHESTRATOR                        │
│         (Spawns/despawns based on demand)                   │
│                            ↓                                 │
├─────────────────────────────────────────────────────────────┤
│  ┌─────────────────────────────────────────────────────┐   │
│  │ GAME SERVERS (Regional, Auto-scaled)                 │   │
│  │ [US-East] [US-West] [EU-West] [Asia] [Oceania]      │   │
│  └─────────────────────────────────────────────────────┘   │
│                            ↓                                 │
├─────────────────────────────────────────────────────────────┤
│                    DATABASE CLUSTER                          │
│  [Player Profiles] [Leaderboards] [Match History] [Items]  │
└─────────────────────────────────────────────────────────────┘

Matchmaking System

MATCHMAKING FLOW:
┌─────────────────────────────────────────────────────────────┐
│  1. QUEUE: Player enters matchmaking queue                  │
│     → Store: skill rating, region, preferences             │
│                                                              │
│  2. SEARCH: Find compatible players                         │
│     → Same region (or expand after timeout)                │
│     → Similar skill (±100 MMR, expand over time)           │
│     → Compatible party sizes                               │
│                                                              │
│  3. MATCH: Form teams when criteria met                     │
│     → Balance teams by total MMR                           │
│     → Check for premade groups                             │
│                                                              │
│  4. PROVISION: Request game server                          │
│     → Orchestrator spawns or assigns server                │
│     → Wait for server ready                                │
│                                                              │
│  5. CONNECT: Send connection info to all players            │
│     → IP:Port or relay token                               │
│     → Timeout if player doesn't connect                    │
└─────────────────────────────────────────────────────────────┘

Player Data Management

// ✅ Production-Ready: Player Session
public class PlayerSession
{
    public string PlayerId { get; }
    public string SessionToken { get; }
    public DateTime CreatedAt { get; }
    public DateTime LastActivity { get; private set; }

    private readonly IDatabase _db;
    private readonly ICache _cache;

    public async Task GetProfile()
    {
        // Try cache first
        var cached = await _cache.GetAsync($"profile:{PlayerId}");
        if (cached != null)
        {
            return cached;
        }

        // Fall back to database
        var profile = await _db.GetPlayerProfile(PlayerId);

        // Cache for 5 minutes
        await _cache.SetAsync($"profile:{PlayerId}", profile, TimeSpan.FromMinutes(5));

        return profile;
    }

    public async Task UpdateStats(MatchResult result)
    {
        LastActivity = DateTime.UtcNow;

        // Update in database
        await _db.UpdatePlayerStats(PlayerId, result);

        // Invalidate cache
        await _cache.DeleteAsync($"profile:{PlayerId}");
    }
}

WebSocket Reconnection Pattern

// ✅ Production-Ready: WebSocket with exponential backoff
class GameWebSocket {
  private ws: WebSocket | null = null;
  private reconnectAttempts = 0;
  private maxReconnectDelay = 30000; // 30 seconds max
  private baseDelay = 1000; // Start at 1 second

  connect(url: string) {
    this.ws = new WebSocket(url);

    this.ws.onopen = () => {
      console.log("Connected to game server");
      this.reconnectAttempts = 0; // Reset on success
    };

    this.ws.onclose = (event) => {
      if (event.code === 1000) return; // Normal closure
      this.scheduleReconnect(url);
    };

    this.ws.onerror = (error) => {
      console.error("WebSocket error:", error);
    };
  }

  private scheduleReconnect(url: string) {
    // Exponential backoff: delay = baseDelay * 2^attempts + jitter
    const delay = Math.min(
      this.baseDelay * Math.pow(2, this.reconnectAttempts),
      this.maxReconnectDelay
    );

    // Add jitter (±20% randomness to prevent thundering herd)
    const jitter = delay * 0.2 * (Math.random() - 0.5);
    const reconnectDelay = delay + jitter;

    console.log(`Reconnecting in ${reconnectDelay}ms (attempt ${this.reconnectAttempts + 1})`);

    setTimeout(() => {
      this.reconnectAttempts++;
      this.connect(url);
    }, reconnectDelay);
  }

  disconnect() {
    if (this.ws) {
      this.ws.close(1000, "Client disconnect");
      this.ws = null;
    }
  }
}

Auto-Scaling Strategy

SCALING TRIGGERS:
┌─────────────────────────────────────────────────────────────┐
│  SCALE UP when:                                              │
│  • Queue time > 60 seconds                                  │
│  • Server utilization > 70%                                 │
│  • Approaching peak hours                                   │
│                                                              │
│  SCALE DOWN when:                                            │
│  • Server utilization < 30% for 15+ minutes                │
│  • Off-peak hours                                           │
│  • Allow graceful drain (don't kill active matches)        │
├─────────────────────────────────────────────────────────────┤
│  PRE-WARMING:                                                │
│  • Spin up servers before expected peak                    │
│  • Use historical data to predict demand                   │
│  • Keep warm pool for instant availability                 │
└─────────────────────────────────────────────────────────────┘

🔧 Troubleshooting

┌─────────────────────────────────────────────────────────────┐
│ PROBLEM: Long matchmaking times                             │
├─────────────────────────────────────────────────────────────┤
│ SOLUTIONS:                                                   │
│ → Expand skill range over time                              │
│ → Allow cross-region matching                               │
│ → Reduce minimum player count                               │
│ → Add bots to fill partial matches                          │
└─────────────────────────────────────────────────────────────┘

┌─────────────────────────────────────────────────────────────┐
│ PROBLEM: Server crashes during peak                         │
├─────────────────────────────────────────────────────────────┤
│ SOLUTIONS:                                                   │
│ → Pre-warm servers before peak                              │
│ → Increase max server instances                             │
│ → Add circuit breakers                                      │
│ → Implement graceful degradation                            │
└─────────────────────────────────────────────────────────────┘

┌─────────────────────────────────────────────────────────────┐
│ PROBLEM: Database bottleneck                                │
├─────────────────────────────────────────────────────────────┤
│ SOLUTIONS:                                                   │
│ → Add caching layer (Redis)                                 │
│ → Use read replicas                                         │
│ → Shard by player ID                                        │
│ → Queue non-critical writes                                 │
└─────────────────────────────────────────────────────────────┘

Infrastructure Costs

| Scale | Players | Servers | Monthly Cost | |-------|---------|---------|--------------| | Small | 1K CCU | 10 | $500-1K | | Medium | 10K CCU | 100 | $5K-10K | | Large | 100K CCU | 1000 | $50K-100K | | Massive | 1M+ CCU | 10000+ | $500K+ |

---

Use this skill: When building online backends, scaling systems, or implementing matchmaking.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: medy-gribkov
• Source: medy-gribkov/arcana
• License: Apache-2.0
• Homepage: https://www.npmjs.com/package/@sporesec/arcana

Install and usage instructions live in the source repository linked above.