Caching Strategy

Server-Side Cache

• 30-second TTL on all paid endpoints
• Identical requests within 30s = FREE (same wallet, same URL, same params)
• Cache key: wallet_address + endpoint + query_params
• Cache header: X-Cache: HIT or X-Cache: MISS

Client-Side Caching

In-Memory Cache (Python)

import time

class CachedClient:
    def __init__(self, api_client, ttl=30):
        self.api = api_client
        self.ttl = ttl
        self.cache = {}

    def get(self, url):
        now = time.time()
        if url in self.cache:
            age = now - self.cache[url]['time']
            if age < self.ttl:
                return self.cache[url]['data']
        data = self.api.get(url).json()
        self.cache[url] = {'data': data, 'time': now}
        return data

Redis Cache (Production)

import redis
import json

cache = redis.Redis(host='localhost', port=6379, db=0)

def get_price(symbol):
    key = f"foursec:price:{symbol}"
    cached = cache.get(key)
    if cached:
        return json.loads(cached)
    data = client.get(f"https://api.foursec.xyz/price/{symbol}").json()
    cache.setex(key, 28, json.dumps(data))  # 28s (under 30s server cache)
    return data

Multi-Layer Cache

Layer 1: In-memory (TTL: 10s) - fastest, single process
Layer 2: Redis (TTL: 25s) - shared across processes
Layer 3: Server cache (TTL: 30s) - free after payment

Cache Invalidation

When to Bypass Cache

• Trading execution (need freshest data)
• Arbitrage detection (stale data = lost profit)

When to Use Cache

• Dashboard display (30s delay acceptable)
• Portfolio valuation (1-minute accuracy fine)
• Historical analysis (data doesn't change)

Cost Optimization

Strategy	Savings	Best For
30s server cache	50-70%	All use cases
60s client cache	80%	Dashboards, trackers
5min client cache	95%	Research, analytics
Batch + cache	90%	Multi-token queries

TIP

Combine client-side caching (60s) with server cache (30s) for maximum savings. First request pays, next 2 are free.