Estratégias de Cache em Memória e Externo
Compare diferentes abordagens de cache, incluindo caches simples em memória e soluções externas robustas, como Redis.
Estratégias de Cache em Memória e Externo é uma aula grátis de LLM Apps in Production (RAG + Vector DB + Caching) no CoddyKit. Esta é a aula 2 de 4. Você pode ler a aula completa abaixo gratuitamente — depois pratica ao vivo no navegador com um editor de código integrado e um tutor de IA 24/7. Faz parte do caminho de aprendizado de LLM Apps in Production (RAG + Vector DB + Caching), e seu progresso é sincronizado entre a web e o app CoddyKit. O curso de LLM Apps in Production (RAG + Vector DB + Caching) inclui 4 aulas no total.
Partes desta aula ainda não foram traduzidas e aparecem em inglês.
Intro to Caching Strategies
Caching is vital for making LLM applications faster and more cost-effective. But not all caches are built the same!
In this lesson, we'll dive into two primary strategies: in-memory caching and external caching. Each has unique benefits and drawbacks depending on your application's needs.
In-Memory Caching: The Basics
In-memory caching means storing data directly within your application's Random Access Memory (RAM). Think of it like a temporary notepad your app keeps handy.
- Speed: Accessing data from RAM is incredibly fast.
- Simplicity: Often easy to set up, using built-in language features (like dictionaries or hash maps).
- No External Dependencies: Your app doesn't need to connect to another service.
Simple In-Memory Cache (Python)
Here's a basic Python example using a dictionary to simulate an in-memory cache for LLM responses. Notice how subsequent requests for the same prompt hit the cache.
cache = {}
def get_llm_response(prompt):
if prompt in cache:
print("Cache hit!")
return cache[prompt]
else:
print("Cache miss, calling LLM...")
# Simulate a slow LLM call
response = f"LLM response for: {prompt}"
cache[prompt] = response
return response
if __name__ == "__main__":
print(get_llm_response("What is RAG?"))
print(get_llm_response("What is RAG?"))
print(get_llm_response("Tell me a joke."))
print(get_llm_response("Tell me a joke."))Limitations of In-Memory Caches
While fast and simple, in-memory caches have significant drawbacks for production LLM applications:
- Ephemeral Data: All cached data is lost if your application restarts or crashes.
- Limited Scale: Each instance of your application has its own separate cache. If you run multiple servers, they won't share data, leading to duplicated work.
- Memory Usage: Large caches can consume a lot of RAM, potentially impacting your application's overall performance.
Introducing External Caching
To overcome the limitations of in-memory caches, we use external caching solutions. These store cached data outside your application, typically in a dedicated server or service.
This allows multiple instances of your application to access and share the same cached data, making it ideal for scalable, distributed systems.
Redis: A Popular External Cache
Redis (Remote Dictionary Server) is a popular open-source, in-memory data store. It's widely used as a cache, database, and message broker due to its high performance and versatile data structures.
It's an excellent choice for external caching in LLM applications because it's incredibly fast and designed for network-based access.
Advantages of External Caching
External caches like Redis provide several key advantages:
- Distributed: Multiple application instances can share a single, consistent cache.
- Persistent: Data can be configured to be saved to disk, so it survives application or cache server restarts.
- Scalable: The cache can be scaled independently of your application, handling massive amounts of data and requests.
- Rich Features: Redis offers advanced features like Time-To-Live (TTL) for automatic cache expiration and various data structures.
Using Redis for LLM Caching (Python)
Here's how you might interact with Redis from Python using the redis-py library. This code assumes a Redis server is running locally on localhost:6379.
It demonstrates setting a key with an expiration (TTL) and retrieving it.
import redis
import json
# Connect to Redis. Ensure a Redis server is running!
# e.g., on Docker: docker run --name my-redis -p 6379:6379 -d redis
try:
r = redis.Redis(host='localhost', port=6379, db=0)
r.ping() # Check connection
print("Connected to Redis successfully!")
except redis.exceptions.ConnectionError as e:
print(f"Could not connect to Redis: {e}")
print("Please ensure a Redis server is running on localhost:6379")
r = None # Set r to None if connection fails
def get_llm_response_from_redis(prompt):
if r is None:
return {"error": "Redis not connected, cannot cache."}
cache_key = f"llm_response:{prompt}"
cached_data = r.get(cache_key)
if cached_data:
print("Redis Cache hit!")
return json.loads(cached_data.decode('utf-8'))
else:
print("Redis Cache miss, calling LLM...")
# Simulate LLM call and create a response structure
response_data = {"text": f"LLM response for: {prompt}", "source": "LLM"}
# Cache the response for 3600 seconds (1 hour)
r.setex(cache_key, 3600, json.dumps(response_data))
return response_data
if __name__ == "__main__":
print(get_llm_response_from_redis("What is the capital of France?"))
print(get_llm_response_from_redis("What is the capital of France?"))
print(get_llm_response_from_redis("Who invented the light bulb?"))
print(get_llm_response_from_redis("Who invented the light bulb?"))Choosing the Right Caching Strategy
Your choice of caching strategy depends on your application's requirements:
- Use In-Memory Caches if:
Your application runs as a single instance, data loss on restart is acceptable, or you're caching very small, temporary datasets. - Use External Caches (e.g., Redis) if:
You need distributed caching across multiple application instances, data persistence is critical, your dataset is large, or you require advanced caching features and scalability.
For most production LLM applications, external caching is the robust choice.
Caching Strategy Quiz
Test your understanding of caching strategies!
Recap: Cache Your Knowledge
We've explored the two main caching strategies for LLM applications:
- In-memory caches are fast and simple but are limited to a single application instance and lose data on restart.
- External caches like Redis offer persistence, distributed sharing, and independent scalability, making them ideal for robust production LLM systems.
Choosing the right strategy depends on your application's scale, data persistence needs, and operational complexity.
Perguntas Frequentes
A aula “Estratégias de Cache em Memória e Externo” é grátis?
Sim — o texto completo de “Estratégias de Cache em Memória e Externo” é grátis para ler aqui na web. Para praticá-la interativamente (um editor de código integrado e um tutor de IA 24/7) e desbloquear o restante do curso de LLM Apps in Production (RAG + Vector DB + Caching), atualize para CoddyKit PRO. O curso de LLM Apps in Production (RAG + Vector DB + Caching) inclui 4 aulas no total.
O que vou aprender em “Estratégias de Cache em Memória e Externo”?
Compare diferentes abordagens de cache, incluindo caches simples em memória e soluções externas robustas, como Redis. Você pratica LLM Apps in Production (RAG + Vector DB + Caching) com código prático que executa diretamente no navegador, e um tutor de IA 24/7 responde suas dúvidas enquanto trabalha na aula.
Preciso ter experiência prévia para começar LLM Apps in Production (RAG + Vector DB + Caching)?
Nenhuma experiência prévia é necessária. LLM Apps in Production (RAG + Vector DB + Caching) no CoddyKit é estruturado para alunos iniciantes até avançados, então você pode começar aqui ou desde o início e aprender no seu ritmo. Esta é a aula 2 de 4.
Quanto tempo leva a aula “Estratégias de Cache em Memória e Externo”?
A maioria das aulas CoddyKit leva cerca de 5–10 minutos. Cada uma é compacta e interativa, então você faz progresso constante e retoma exatamente de onde parou entre web e app.
Posso escrever e executar código nesta aula de LLM Apps in Production (RAG + Vector DB + Caching)?
Sim. Cada aula de LLM Apps in Production (RAG + Vector DB + Caching) inclui um editor de código integrado, então você escreve e executa código real direto no navegador e recebe feedback de IA instantaneamente — nenhuma configuração local necessária.
Todas as aulas deste curso
- A Importância do Armazenamento em Cache de Chamadas a LLMs
- Estratégias de Cache em Memória e Externo
- Integrando Cache a um Pipeline RAG
- Cache semântico para aplicativos de LLM