Glossary
Horizontal Scaling
Increasing AI agent system capacity by deploying additional parallel instances rather than upgrading individual components.
What is Horizontal Scaling?
Horizontal scaling adds more agent instances to handle increased load, contrasting with vertical scaling that upgrades existing instances with more resources. This approach provides better fault tolerance since no single instance failure impacts overall capacity, and enables near-linear scaling by simply adding instances. Stateless agent design facilitates horizontal scaling by allowing any instance to handle any request.
Effective horizontal scaling requires load balancing to distribute work evenly, session management if agents maintain state, and orchestration to automatically adjust instance count based on demand. Cloud deployments make horizontal scaling straightforward through auto-scaling groups. Challenges include coordinating shared resources like databases and managing costs when over-provisioned.
Example
A customer service agent system normally runs 10 instances handling 500 requests per minute. During a product launch, traffic spikes to 2,000 requests per minute. Auto-scaling detects the load increase and deploys 30 additional instances within two minutes, maintaining response times throughout the spike.
How Signet addresses this
Signet's Reliability dimension values horizontal scaling as evidence of mature operational design. Agents with demonstrated auto-scaling maintaining performance during load spikes score higher in reliability than those with fixed capacity prone to overload.
Build trust into your agents
Register your agents with Signet to receive a permanent identity and trust score.