Corewave

What is Cubewave?

Cubewave serves as the Heroku for AI agents, allowing easy deployment of your LLM in a secure cloud environment from tools like Rivet, langflow...

Cubewave is providing tools to make deploying microVMs through Firecracker simpler, minimizing the complexities entailed in the process.

• Corewave operates as the equivalent of "docker-compose" for microVMs, enabling the configuration and creation of microVMs through a YAML file.

Community and Updates

To keep updated on our progress, join our discord server and follow us on Twitter!

Requirements:

KVM requirements:

In order to create micro vm, you need to make sure KVM is enabled on your machine

lsmod | grep kvm

The output should be something like this:

kvm_intel             342150  0
kvm                   960692  1 kvm_intel
irqbypass              15324  1 kvm

OS requirements:

Cubewave currently runs on Linux Ubuntu and you also need to have at least 5GB of available storage

Corewave:

Installation

    wget --quiet -O - https://repo.cubewave.dev/cubewave-pubkey.asc | sudo tee /etc/apt/keyrings/cubewave-pubkey.asc 
    echo "deb [signed-by=/etc/apt/keyrings/cubewave-pubkey.asc arch=$( dpkg --print-architecture )] https://repo.cubewave.dev $(lsb_release -cs) main" | sudo tee /etc/apt/sources.list.d/apt-cubewave.list
    apt update
    apt install corewave

To verify that corewave is successfully installed run corewave version

you should get something like this: corewave v.0.1

Running your first MicroVM

YAML file structure:

Corewave requires a YAML files that defines the configuration of the MicroVM(s) you want to deploy, below is how the file should be structured:

version: "1"
specs:
  MyApp1:
    logsPath: "FC_MyApp.log"
    CPU: 1
    RAM: 1024
    image: "sample_app"

  MyApp2:
    ...

  MyApp3:
    ...

version: Configuration version

spec: Specs block, within we specify the configuration of all our MicroVMs

MyApp1: Name of the MicroVM to provision (you can name it whatever you want)

logsPath: Path of the Firecracker log file

CPU: Number of host CPUs to allocate to the MicroVM

RAM: Amount of RAM to allocate to the MicroVM (in MB)

Image(optional): Absolute path to any executable you want to run as a daemon in the MicroVm.

Running Corewave

Once your YAML file is ready you run Corewave and pass it the configuration file using the -f flag:

sudo corewave -f config.yml

Corewave will then generate the necessary infrastructure components on your host VM and run firecracker with the required configuration to provision your MicroVMs.

Note: For now Corewave needs to run with privileged permissions so please make sure you are either root or using sudo when running Corewave

/!\ After each new executing of Corewave to provision new MicroVMs, the previous set of MicroVMs and data will be deleted. This will be changed in the upcoming release. /!\

Working with MicroVm:

Once your MicroVMs are up and running, Corewave provides a set of commands to easily monitor and interact with your MicroVMs:

MicroVM data and configuration files:

Corewave stores all data and configuration related to each MicroVM in the default $HOME directory under .cubewave

Listing MicroVM:

Use corewave ps to get the list of currently running MicroVMs

You can also use the ps argument to get information about a specific MicroVM only, by providing its id using corewave ps {id}

Connecting to a MicroVM:

You can simply use ssh to connect to a MicroVM, authentication is based on an ssh key that located in the config directory of every MicroVM under $HOME/.cubewave/{ID}/fs/ubuntu-22.04.id_rsa

{ID} is ID of the MicroVM you want to connect to.