Memory on demand

What it is

Take memory back from a running VM and give it back again, live.

FlashVMM reclaims memory from a running FlashVM through the balloon and returns it on demand. The guest keeps running the whole time. Idle VMs cost the host almost nothing.

Why it matters

Memory is the cost of running many sandboxes. FlashVMM spends memory where the work is.

• Tiny footprint. An idle VM uses as little as 27 MB of host memory.
• Live reclaim. Memory taken back while the VM runs.
• Instant return. Give it back the moment the VM needs it.

Runs on

• Linux x86_64
• Linux arm64
• macOS
• Windows x64
• Windows arm64

How it works

Set a balloon target in pages; zero returns everything.

CLI

flash balloon vm1 --target-pages 16384
flash exec vm1 -- free -m
flash balloon vm1 --target-pages 0

MCP

flash_cli(args: ["balloon", "vm1", "--target-pages", "16384"])
flash_resource(vm_id: "vm1", resource: "balloon")

API

PATCH /v1/flashvm/memory/balloon {"target_pages": 16384}
GET /v1/flashvm/memory/balloon

The result

Memory taken back from a running VM and returned on demand, with the guest running throughout.

Dense hosts, with memory flowing to the VMs that need it.