How do containers provide storage on demand basis?
Depending on the workload, how are the number of containers to be deployed decided?
Hi @sanjana.
This is such a open question, there are to many factor and variables that come into the game. But I'll try to clarify you a as much as I can.
Docker containers by itself don't dynamically scale on demand, you need an underlying service that orchestrates autoscaling based on different factors as can be disk space, CPU or memory load, etc.
Number of containers to deploy also depends of what storage backend you want to use. EG: With ceph you have minimum requirements for OSD, MONs, etc.
If you are using Ceph (as probably do using the tag kolla in the question), you need a minimum of 3 MON containers, this is to maintain a master-slave quorum. OSD does not have any minimum number of containers (just theory, at least 3 should be used to maintain the cluster in a healthy state and ensure data redundancy).
Containers orchestration, there are some underlay services that can do the orchestration, but containers need to be deployed through them. This services includes: Kubernetes, Docker Swarm, Apache Messos, also Heat if you deploy OpenStack on top of OpenStack).
At this moment, Kolla community are developing an orchestration/deployment layer based on Kubernetes [0]
Hope I clarify some of your doubts.
[0] [http://docs.openstack.org/developer/k...
Regards, Eduardo Gonzalez
Asked: 2016-07-25 05:34:34 -0500
Seen: 96 times
Last updated: Jul 28 '16
Please, could you elaborate a little more the question and what exactly what you want to know?
I am interested in knowing how containers scale dynamically based on demand. What are the existing software that support this and how exactly it is being done.
I answered your question, if you need something else to clarify, edit your question and i'll try to solve it. Regards