Kubernetes Persistent Volumes and Claims

In a distrubuted system, storage isn't simple to manage without a good abstraction, kubernetes offer an object called persistent volume to let pod to reclaim storage for persisting data. In the pod definition you can define your volume type with another object called persistentVolumeClaim where you can define type size and many other things.

There are different phase in the persistent storage lifecycle:
  • Provisioning: when an administrator create a Persistent Volume or a cloud resource dynamically request
  • Binding: when a PVC is created  k8s control loop watch if there are Persisten Volume matching the PVC  and bind them together.
  • Use: when the bound volume is mounted in to the pod
  • Releasing: whene the POD is done and an API request is sent to delete the related bound PVC, the volume in this stage remains
  • Reclaim: there are three options related to this state:
    -Retain: keep data interact and allow an administrato to handle data from storage
    -Delete: tells the volume to delete the API object and the storage behind.
    -Recycle: runs an rm -rf /mountpint and volume become availble to a new claim.
As you can understand everything in kubernetes is designed to be available during scaling or pod destroy in this way.

Comments

Post a Comment

Popular posts from this blog

IBM and Red Hat long story, long love, now married!

Most of the people thinking about opensource they think in my perspective mostly to Red Hat and other company rather than IBM. IBM is really active in opensource for a long time, helped Apache Foundation , Eclipse Foudation collaborate with the Linux Foundation and Red Hat as well. I think I can create a different long post to describe how is active and how helped opensource in different way. We can analyze  this acquisition from different perspective, from Financial to  Development. My personal point of view ( from a Cloud Architect ) most of the application around the world in banking, assurance and not only are java based, think about java in cloud in the past Year could be hard. Right now we can talk about java in the cloud with opensource project like OpenJ9 and OpenLiberty and the community behind. If you think about moving from on premise to cloud or from monolitch to microservice you can't say " well it's time to rework " because you are throwing years o
Read more

MY EXPERIENCE @ HYBRID CLOUD SUMMIT IN ZURICH

Image
I can simply say “it was a great experience” and based on the audience probably people can trust me, but let's go deeper into the event. My favorite part of the event was the panel session with cloud discussions and question answer about slides and example presented during the session. Based on discussion the most relevant general questions are: Adoption: best practices on adopting cloud technologies Development: how to develop cloud applications in the right way and how we can enable fresh people into this great world IBM and Forrester gave a lot of answers because there are different aspects and factors we need to cover when we talk about cloud development and adoption, let me share some of these: Security: 12 factor principle is the main manifesto to follow Digital Experience: in cloud you need to create applications user centric Performance: in cloud you need to give best experience to the end user in every part of the world Microservice: rework of the legacy
Read more