Você está visualizando o site de uma EDIÇÃO PASSADA do QCon.

Palestra: Multi-tenancy in Kubernetes

Track: Atingindo Todo o Potencial de Containers

Sala: 4 São Francisco

Horário: 11:50am - 12:35pm

Dia da semana: Quarta-feira

Slides: Download Slides

Nível: Intermediário

Persona: Desenvolvedor(a) Programador(a), Desenvolvedor(a) Sênior, DevOps, Gerente de Operações, Gestão (VP, CTO, CIO, Diretoria), Líder Técnico(a)

Apresentação em Inglês

Share this on:

This presentation is now available to view on InfoQ.com

Assista a palestra com transcrição

Pontos Principais

  • Shared clusters can help companies use their kubernetes deployments cost-effectively and reduce operational overhead;
  • Cluster sharing brings with it the need for additional insights tooling (e.g., metrics breakdown by namespace/label, usage metering per namespace/label);
  • The kubernetes model allows for different setups that are best for different situations: isolated clusters, isolated nodes (i.e., each node is dedicated to one tenant), or fully shared clusters. This flexibility within the same model can help drive up operational efficiency.

Resumo

Most business need to run multiple applications (e.g., multiple microservices) or serve many clients (e.g., a SaaS providing a database per client). Running multiple applications with kubernetes, or serving multiple tenants, can be approached in several ways. The simplest, but least cost-effective, is to have a separate cluster for each application or tenant. On the other end of the extreme are shared clusters, where each application or tenant is assigned a namespace (or label) in the kubernetes cluster. The latter model allows for better cost efficiency. In this talk, I will share both the mechanics and the implications of cluster sharing on cost, isolation, and operational efficiency. I will discuss use cases where cluster sharing is appropriate, for instance in the case of multiple small applications, or applications that run at non-overlapping times (e.g., interspersing batch workloads with services when traffic to those services is low). I will also talk about times when cluster sharing is more challenging. At the end of this talk, I hope to give you a good idea for how to think about the different approaches to sharing.

Palestrante: Katharina Probst

Senior Engineering Leader, Kubernetes & SaaS at Google

Katharina Probst is a Senior Engineering Leader, Kubernetes & SaaS at Google. Before this, she was leading engineering teams at Netflix, being responsible for the Netflix API, which helps bring Netflix streaming to millions of people around the world. Prior to joining Netflix, she was in the cloud computing team at Google, where she saw cloud computing from the provider side. Her interests include scalable, distributed systems, APIs, cloud computing, and building effective and successful teams. She also holds a PhD in Computer Science from Carnegie Mellon University.

Find Katharina Probst at

Tracks 2019