Load Balancer Tools
Distributing incoming traffic across multiple backend instances.
Directly related. Edge hosts route to healthy replicas over the mesh, with routing, draining, ejected and pending states, ejection of hung replicas and make-before-break rolling updates.
What load balancing solves
Once an application needs more capacity than one instance can serve, or needs to survive one instance crashing without downtime, something has to sit in front and spread requests across multiple running copies -- round-robin (rotate evenly), least-connections (send to whichever backend is least busy right now), or IP-hash (same client always hits the same backend, useful for sticky sessions) are the common strategies. Health checks matter as much as the distribution algorithm: a load balancer that keeps sending traffic to a backend that's actually down isn't doing its job.
Distributing incoming traffic across multiple backend instances. Directly related. Edge hosts route to healthy replicas over the mesh, with routing, draining, ejected and pending states, ejection of hung replicas and make-before-break rolling updates.