"Infrastructure in the Era of Big Models" Machine Learning Application Development and Operation Platform

ltaodream

"Infrastructure in the Era of Big Models" Machine Learning Application Development and Operation Platform [Copy link]

 

This is the eleventh chapter of the book. In summary, this chapter only summarizes the content involved. For more details, individuals need to continue to learn knowledge and accumulate work experience on a daily basis.

Microservices Platform

• Use Kubernetes to provide basic service registration, discovery and service routing capabilities, implemented through Service
• Use istio to achieve non-intrusive, language-independent service governance and link tracking
• Container platform scheduling
• Splitting Microservices "Infrastructure in the Era of Big Models" Chapter 11 Problems Beyond Algorithms in Machine Learning Applications

Middleware Services

Message Queue Middleware

The book mainly talks about Kafka, RabbitMq, RocketMq

Application Scenario

• Asynchronous optimization performance
• Decoupling between systems
• Traffic peak shaving

Selection inspection points

• reliability
• Commonly used MQ in the industry
• Clustering support
• QPS - Throughput
• Are common functions complete?

contrast

Cache middleware

Redis

For more details, please see several of my other articles, which are more detailed than this part in the book:

• A thorough understanding of Redis source code (data structure)
• A thorough understanding of Redis source code (event-driven framework)

Database middleware

This section mainly talks about the ACID of the database

• Atomicity
• consistency
• Isolation
• Persistence