Main repository for Master thesis on serverless.
- In functions directory there exist 4 versions of the same ML Workflow
- Each version consists of functions of different granularity
- Version1 is the most fine-grained, while Version4 is the most coarse-grained of the versions
- After deploying all the functions, by using the exec script we can run the desired instance on our local cluster as many times as we like in a sequential way of invocations
./exec.sh version1 url1 --times=10 yes yes --sleep=5s --step=15s
- We can scale each function's replicas up or down so as to make the versions capable of serving more or less requests per minute
kubectl scale deployment version1 -n openfaas-fn --replicas=2