Fabio Mora
La Unix Way vista da un DevOps.
"Con l'acronimo KISS (Keep It Simple and Stupid) si evoca un modo di fare software antico: la UNIX Way. Ma non è tutto racchiuso lì dentro, si tratta piuttosto di una filosofia progenitrice del movimento DevOps. Declinata oggi nel mondo Linux da migliaia di tecnici e organizzazioni, che include assunti come modularità, semplicità, composizione e chiarezza nel fare. Tutto evolve portandosi dietro principi solidi e al tempo stesso pronti al cambiare: dai modelli open-source ai flussi Git, dalla gestione dei pacchetti ai container, dalle infrastrutture a risorse finite agli orchestratori del cloud; fino a mescolare le definizioni stesse di sistema operativo e applicazione tra user-space e kernel-space. Henry Spencer, padre delle RegEx, scrisse coloro che non capiscono Unix sono condannati a inventarlo di nuovo. In questo talk racconto ciò che ho imparato da questa filosofia che già in tempi non sospetti si occupava di sistemi complessi, e ha ancora molto da regalare a chi è coinvolto nella produzione e nel mantenimento di software."
Giuseppa Muscianisi
Containerization of parallel MPI and accelerated applications on HPC systems
Using containers in Linux has become a common practice in many business environments. Technologies like Docker or Kubernetes, allow to create self-contained environments or ''containers'' that are designed for a specific purpose. Unfortunately, the use of these technologies in High Performance Computing (HPC) environment has been problematic due to security concerns. The open source project Singularity addresses this problem, allowing for the creation of containers that maintain the security infrastructure of the machine they are running on. After a general introduction of Singularity, two containerized applications are presented: 1) Quantum Espresso, the integrated suite of Open-Source computer codes for electronic-structure calculations and materials modelling at the nanoscale, and 2) Tensorflow, the open-souce software library for machine learning. The performance comparison between bare metal and container executions is provided, showing a negligible overhead introduction in the containerized version of the applications. All the tests have been run on the supercomputing clusters installed in CINECA.
Paolo Mainardi
The path to a Serverless-native era with Kubernetes
"In questo talk analizzeremo come il paradigma Serverless sta cambiando il modo di sviluppare il design delle applicazioni, incoraggiando un architettura a (micro) servizi, ragionando su un modello FaaS (Functions as a service). Kubernetes grazie ad alcuni framework come Knative, Openfaas, Kubeless, permette dunque di implementare questo modello sulla propria infrastruttura. Vedremo ogni aspetto, dalla security, all'observability fino alle best practices. "
Sascha Grunert
Kubeflow - Data Science on Steroids
"Artificial intelligence and machine learning are the most hyped topics in software development these days! The major goal is to get an understandable output from a huge set of unmanageable input data. Standard de facto frameworks TensorFlow and PyTorch established a rich set of features over time by being well maintained under the hood, but they they’re not enough to address challenges in terms of CI/CD. Kubernetes is emerging as the new, easier, build and test infrastructure in this arena. In this talk we will discover the possibilities to utilize the joint forces of Kubernetes and Data Science. For this we have chosen to elaborate Kubeflow, one of the most common open source cloud native platforms for machine learning. Kubeflow is a project which is dedicated to making deployments of machine learning workflows on Kubernetes simple, portable and scalable. Join us and we will discover the exciting features of Kubeflow together, like spawning and managing Jupyter servers and creating our very own machine learning pipelines from scratch"
