Application Scaling
Hubfly space supports two practical scaling models: Vertical Auto-Scale (dynamically adjusting CPU and memory within defined limits) and Manual Replicas (creating multiple identical container instances for horizontal throughput).
Scaling Model Comparison
| Scaling Model | What It Modifies | Best Workload Types |
|---|---|---|
| Vertical Auto-Scale | Dynamically scales CPU & RAM between a base minimum and maximum limit. | Single-process web apps, CPU-bursty APIs, memory-heavy workloads, admin tools. |
| Load-Balanced Replicas | Creates multiple container copies behind a load balancer group. | Stateless web apps, high-concurrency REST APIs, public frontends. |
| Internal Replicas | Creates multiple container copies without public load balancing. | Background queue workers, job processors, internal batch microservices. |
Vertical Auto-Scale Configuration
Configure vertical auto-scaling by setting a base resource floor and a peak capacity ceiling:
// Vertical Auto-Scale Config Example
Base CPU: 1.0 vCPU · Max CPU: 4.0 vCPU
Base RAM: 1024 MB · Max RAM: 4096 MB
Manual Replicas Workflow
Use the Replicate flow on any container to spawn 1 to 20 additional identical instances.
- Scale Replicas + Load Balancer: Hubfly space automatically adds new replicas to an existing or new Load Balancer group for horizontal HTTP traffic distribution.
- Internal Replicas: Spawns same-app containers connected to the private project network for worker pools pulling from a shared queue (e.g. Redis or RabbitMQ).