Fission is a fast serverless framework for Kubernetes with a focus on developer productivity and high performance.
Fission operates on just the code: Docker and Kubernetes are abstracted away under normal operation, though you can use both to extend Fission if you want to.
Fission is extensible to any language; the core is written in Go, and language-specific parts are isolated in something called environments (more below). Fission currently supports NodeJS and Python, with more languages coming soon.
Fission maintains a pool of "warm" containers that each contain a small dynamic loader. When a function is first called, i.e. "cold-started", a running container is chosen and the function is loaded. This pool is what makes Fission fast: cold-start latencies are typically about 100msec.
We're built on Kubernetes because we think any non-trivial app will use a combination of serverless functions and more conventional microservices, and Kubernetes is a great framework to bring these together seamlessly.
Building on Kubernetes also means that anything you do for operations on your Kubernetes cluster — such as monitoring or log aggregation — also helps with ops on your Fission deployment.
A function is a piece of code that follows the fission function interface.
An environment contains the language- and runtime-specific parts of running a function. Fission comes with NodeJS and Python environments; you can also extend environments or create entirely new ones if you want. (An environment is essentially just a container with a webserver and dynamic loader.)
A trigger is something that maps an event to a function; Fission supports HTTP routes as triggers today, with upcoming support for other types of event triggers, such as timers and Kubernetes events.
# Add the stock NodeJS env to your Fission deployment
$ fission env create --name nodejs --image fission/node-env
# A javascript one-liner that prints "hello world"
$ echo 'module.exports = function(context, callback) { callback(200, "Hello, world!\n"); }' > hello.js
# Upload your function code to fission
$ fission function create --name hello --env nodejs --code hello.js
# Map GET /hello to your new function
$ fission route create --method GET --url /hello --function hello
# Run the function. This takes about 100msec the first time.
$ curl http://$FISSION_ROUTER/hello
Hello, world!You can install Kubernetes on your laptop with minikube:
$ curl -LO https://storage.googleapis.com/kubernetes-release/release/$(curl -s https://storage.googleapis.com/kubernetes-release/release/stable.txt)/bin/darwin/amd64/kubectl && chmod +x kubectl && sudo mv kubectl /usr/local/bin
$ curl -Lo minikube https://storage.googleapis.com/minikube/releases/v0.14.0/minikube-darwin-amd64 && chmod +x minikube && sudo mv minikube /usr/local/bin/
$ minikube start $ curl -LO https://storage.googleapis.com/kubernetes-release/release/$(curl -s https://storage.googleapis.com/kubernetes-release/release/stable.txt)/bin/linux/amd64/kubectl && chmod +x kubectl && sudo mv kubectl /usr/local/bin
$ curl -Lo minikube https://storage.googleapis.com/minikube/releases/v0.14.0/minikube-linux-amd64 && chmod +x minikube && sudo mv minikube /usr/local/bin/
$ minikube startOr, you can use Google Container Engine's free trial to get a 3 node cluster.
$ kubectl version
If you're using minikube or no cloud provider, use these commands to set up services with NodePort. This exposes fission on ports 31313 and 31314.
$ kubectl create -f http://fission.io/fission.yaml
$ kubectl create -f http://fission.io/fission-nodeport.yaml
Set the FISSION_URL and FISSION_ROUTER environment variables.
FISSION_URL is used by the fission CLI to find the server.
FISSION_URL should be prefixed with a http://. (FISSION_ROUTER is
only needed for the examples below to work.)
If you're using minikube, use these commands:
$ export FISSION_URL=http://$(minikube ip):31313
$ export FISSION_ROUTER=$(minikube ip):31314
If you're using GKE or any other cloud provider that supports the LoadBalancer service type, use these commands:
$ kubectl create -f http://fission.io/fission.yaml
$ kubectl create -f http://fission.io/fission-cloud.yaml
Save the external IP addresses of controller and router services in
FISSION_URL and FISSION_ROUTER, respectively. Wait for services to
get IP addresses (check this with kubectl --namespace fission get svc). Then:
$ export FISSION_URL=http://$(kubectl --namespace fission get svc controller -o=jsonpath='{..ip}')
$ export FISSION_ROUTER=$(kubectl --namespace fission get svc router -o=jsonpath='{..ip}')
Get the CLI binary for Mac:
$ curl http://fission.io/mac/fission > fission && chmod +x fission && sudo mv fission /usr/local/bin/
Or Linux:
$ curl http://fission.io/linux/fission > fission && chmod +x fission && sudo mv fission /usr/local/bin/
Finally, you're ready to use Fission!
$ fission env create --name nodejs --image fission/node-env
$ echo 'module.exports = function(context, callback) { callback(200, "Hello, world!\n"); }' > hello.js
$ fission function create --name hello --env nodejs --code hello.js
$ fission route create --method GET --url /hello --function hello
$ curl http://$FISSION_ROUTER/hello
Hello, world!
[You only need to do this if you're making Fission changes; if you're just deploying Fission, use fission.yaml which points to prebuilt images.]
You'll need go installed, along with the glide dependecy management tool. You'll also need docker for building images.
The server side is compiled as one binary ("fission-bundle") which contains controller, poolmgr and router; it invokes the right one based on command-line arguments.
To build fission-bundle: clone this repo, then from the top level directory:
# Get dependencies
$ glide install
# Build fission server and an image
$ pushd fission-bundle
$ ./build.sh
# Edit push.sh to point to your registry
$ $EDITOR push.sh
$ ./push.sh
$ popd
# To install, update fission.yaml to point to your compiled image
$ $EDITOR fission.yaml
$ kubectl create -f fission.yaml
If you're changing the CLI:
# Build Fission CLI
$ cd fission && go install
Fission is in early alpha. Don't use it in production just yet!
Right now, we're looking for developer feedback -- tell us which languages you care about, what use cases you might use it for, and so on.