[KongchangAI]
Concept横向扩展 / Horizontal Scaling

Horizontal Scaling

A scaling approach that increases a system's processing capacity by adding more server instances. It is contrasted with vertical scaling and is the foundation of stateless architectures and large-scale distributed systems.

Timeline (last 90 days)

Nov 10

Horizontal expansion refers to that, after semantic retrieval hits initial seed memories, the system expands along the graph structure to their neighboring nodes, bringing semantically related memories that were not directly retrieved into the candidate set.

Unverified50%
Nov 10

Horizontal expansion broadens the recall coverage of memories, while vertical composition optimizes how each memory's evidence contributes, and the two work in concert.

Unverified50%
Nov 5

The key benefit of externalizing session state is making the Agent process stateless, thereby supporting horizontal scaling, rolling releases and failover—any instance can restore context from storage based on the session ID.

Unverified50%
Sep 15

Horizontal scaling (read-write splitting, sharding) requires coordinated modifications at the application layer, making it more complex than vertical scaling.

Unverified50%

All Facts (4)

Source Articles