Skip to content
This repository has been archived by the owner on Aug 19, 2021. It is now read-only.
/ faasm-experiments Public archive

Experiments and transient code related to Faasm

Notifications You must be signed in to change notification settings

faasm/faasm-experiments

Repository files navigation

ARCHIVAL

This repo is no longer used, and most of the experiments are not maintained.

For more recent experiments see this repo.

If trying to resurrect experiments in here, be aware that they use various utilities since removed from Faasm in this PR

Faasm experiments

This repository holds all experimental/ transient code related to Faasm.

Docs on specific experiments are held in the relevant subfolders.

This code assumes you have Faasm checked out at /usr/local/code/faasm, and it is symlinked as such from third-party/faasm. If you have Faasm checked out elsewhere, just change the symlink.

Local Set-up

If building and these experiments locally, you should set up Faasm for local development.

You must then make sure WAVM and WAMR are up to date:

cd third-party/faasm
git submodule update --init third-party/WAVM
git submodule update --init third-party/wamr

Finally, to use the Faasm CLI, you need to run:

# Set up the Faasm CLI
source third-party/faasm/workon.sh

# Check 
inv toolchain.version

Remote Set-up

If you're running experiments from a remote client somewhere, you can set it up by creating an inventory file at ansible/inventory/benchmark.yml that looks like:

[all]
<client_host>

You then need to SSH onto the host and make sure Faasm is checked out at /usr/local/code/faasm:

Then intall the relevant code and deps with:

cd ansible
ansible-playbook -i inventory/benchmark.yml experiment_client.yml

Microbenchmarks Set-up

To execute the microbenchmarks you'll need to run a few more steps on your experiment host. First, SSH in and run:

./bin/set_up_benchmarks.sh

Then download the toolchain, sysroot and runtime root:

inv toolchain.download-toolchain
inv toolchain.download-sysroot
inv toolchain.download-runtime

Python

From your client machine (local or remote), you'll need to run Python tasks. This project adds extra tasks to the existing Faasm CLI.

You can either set up a new virtual env for this project, or use the existing Faasm CLI environment from your Faasm checkout (via source workon.sh).

You then need to make sure all the python deps are installed:

cd third-party/faasm/faasmcli
pip install -r requirements.txt 
pip install -e .

# Check it's all working
cd ../../..
inv -l

Knative Experiments

Experiments that are run on a Kubernetes cluster must be run from a machine with access to the cluster that can run kubectl and kn.

Billing Estimates

To get resource measurements from the hosts running experiments we need an inventory file that includes all the hosts we want to measure at ansible/inventory/billing.ini, something like:

[all]
myhost1
myhost2
...

Then we can run the set-up with:

cd ansible
ansible-playbook -i inventory/billing.ini billing_setup.yml

About

Experiments and transient code related to Faasm

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •