From the course: AI Solution Design Patterns: Data, Model Training, and Application Architectures
Join today to access over 24,300 courses taught by industry experts.
Distributed AI processing
From the course: AI Solution Design Patterns: Data, Model Training, and Application Architectures
Distributed AI processing
- The distributed AI processing pattern simply addresses the problem of limiting an AI system's deployment to a centralized environment. Many AI solutions by default are installed and set up in a central deployment environment, like a single server. During both training and production stages, this can severely inhibit the AI system's performance and can also lead to runtime failure conditions when the capacity of the underlying infrastructure is exceeded. The biggest problem with this type of situation is that once it occurs, it can be impactful to introduce new infrastructure, especially if the AI system is already in production. The distributed AI processing pattern provides the option of having AI solution components hosted by a collection of servers. This allows the AI solution's workload to be shared across multiple servers to avoid performance bottlenecks and potential failure conditions. There are generally two ways that this pattern can be applied. There is fine-grained…
Download courses and learn on the go
Watch courses on your mobile device without an internet connection. Download courses using your iOS or Android LinkedIn Learning app.