Agent Substrate
A runtime environment platform designed for agent workloads. It abstracts the agent runtime environment as an Actor, runs on top of Workers, and is managed uniformly by a WorkerPool, supporting suspension and on-demand scheduling.
Timeline (last 90 days)
Through the suspend-wake mechanism and HPA-driven Worker scaling, Agent Substrate shifts the scaling strategy from pre-allocating large amounts of resources to dynamically scaling according to actual scheduling demand, in order to reduce resource waste and guarantee response speed.
Agent Substrate abstracts the Agent's runtime environment as an Actor; the Actor runs on top of a Worker, and the Worker is managed uniformly by a WorkerPool.
In Agent Substrate, an Actor is in a suspended state when not scheduled, existing only as a record and consuming no CPU or memory.
Each Worker has a memory limit of 1Gi, while each Actor template requires 768Mi, so one Worker can only hold one Actor.
In Agent Substrate, Actors can be created in bulk via the `kubectl ate create actor` command, and after creation the initial state is ACTOR_STATE_SUSPENDED.
In Agent Substrate, after two Workers are each filled by one Actor (768Mi/1Gi), requesting a third Actor returns a 504 timeout, because there is not enough free memory to schedule it.
After manually suspending a running Actor to free memory, the waiting Actor can then be scheduled successfully.
In Agent Substrate, the WorkerPool can be configured as the scaling target of HPA, scheduling pressure (the number of Actors queued waiting for memory) becomes the signal that triggers scaling, and the WorkerPool can scale up to 10 replicas.
After the WorkerPool scaled from 2 replicas to 6 replicas, all six awakened Actors were scheduled and the request returned 200.
This article chose HPA rather than VPA because the memory limit of a single Worker determines that it can only hold a fixed number of Actors; enlarging a single Worker's memory cannot solve the scheduling bottleneck, and increasing the number of Workers is the correct scaling direction.
9 more timeline events
All Facts (18)
Through the suspend-wake mechanism and HPA-driven Worker scaling, Agent Substrate shifts the scaling strategy from pre-allocating large amounts of resources to dynamically scaling according to actual scheduling demand, in order to reduce resource waste and guarantee response speed.
50%UnverifiedIn Agent Substrate, an Actor is in a suspended state when not scheduled, existing only as a record and consuming no CPU or memory.
50%UnverifiedEach Worker has a memory limit of 1Gi, while each Actor template requires 768Mi, so one Worker can only hold one Actor.
50%UnverifiedIn Agent Substrate, Actors can be created in bulk via the `kubectl ate create actor` command, and after creation the initial state is ACTOR_STATE_SUSPENDED.
50%UnverifiedIn Agent Substrate, after two Workers are each filled by one Actor (768Mi/1Gi), requesting a third Actor returns a 504 timeout, because there is not enough free memory to schedule it.
50%UnverifiedAfter manually suspending a running Actor to free memory, the waiting Actor can then be scheduled successfully.
50%UnverifiedIn Agent Substrate, the WorkerPool can be configured as the scaling target of HPA, scheduling pressure (the number of Actors queued waiting for memory) becomes the signal that triggers scaling, and the WorkerPool can scale up to 10 replicas.
50%UnverifiedAfter the WorkerPool scaled from 2 replicas to 6 replicas, all six awakened Actors were scheduled and the request returned 200.
50%UnverifiedThis article chose HPA rather than VPA because the memory limit of a single Worker determines that it can only hold a fixed number of Actors; enlarging a single Worker's memory cannot solve the scheduling bottleneck, and increasing the number of Workers is the correct scaling direction.
50%UnverifiedAgent Substrate abstracts the Agent's runtime environment as an Actor; the Actor runs on top of a Worker, and the Worker is managed uniformly by a WorkerPool.
50%UnverifiedAX's underlying layer is Agent Substrate, a new abstraction layer built on top of Kubernetes.
50%UnverifiedGoogle claims Agent Substrate is designed for hundreds of millions of registered Agents, millions of sub-second tool calls per second, and billions of concurrent sessions per cluster.
50%UnverifiedAgent Substrate replaces heavyweight pods with lightweight actors, packing massive numbers of sessions into ordinary clusters through dense multiplexing.
50%UnverifiedAgent Substrate与LangChain等上层框架更可能形成互补而非直接竞争关系
50%UnverifiedAgent Substrate是一个采用Go语言开发的AI Agent底层运行时项目，官方定位为「the core system」
50%UnverifiedSubstrate并不与LangChain等上层框架直接竞争，而是形成互补关系，上层框架负责开发者体验与业务逻辑编排，Substrate负责底层高性能执行与系统级保障
50%UnverifiedSubstrate的文档完善度、生态成熟度以及社区规模相比头部Python框架仍有差距，处于早期成长阶段
50%Unverified一个成熟的Agent基质层通常需要具备任务生命周期管理、状态持久化、资源隔离与调度、可观测性四大核心能力
50%