newbie questions about ubuntu cloud, maas and storage
I'd like to take OpenStack for a test drive with some spare gear so I've been trying to find and read documentation on it but I'm awash in package names and feel like I'm trying to drink from a firehose. :-)
What I'm shooting for is setting up a private cloud that's got some redundancy where I can lose any single node without losing the VMs running on the cloud. As I understand it, there's some storage mechanism OpenStack uses called Swift that accomplishes this? Does MAAS use swift by default? Is each MAAS node providing storage for the cloud as a whole?
The Ubuntu cloud docs I read say the minimum number of MAAS nodes is 6. But other OpenStack docs seem to imply that one can run with fewer nodes. What is the minimum number of MAAS nodes do I need to have at least minimal redundancy? And what do I need to do besides installing MAAS nodes, a cluster controller and a region controller and juju? Are there extra steps on top of that to have various levels of HA?
All I'm shooting for at this point is storage redundancy and having VMs (at worst) reboot if the node they had been running on died...
I'd like to take OpenStack for a test drive with some spare gear so I've been trying to find and read documentation on it but I'm awash in package names and feel like I'm trying to drink from a firehose. :-)
What I'm shooting for is setting up a private cloud that's got some redundancy where I can lose any single node without losing the VMs running on the cloud. As I understand it, there's some storage mechanism OpenStack uses called Swift that accomplishes this? Does MAAS use swift by default? Is each MAAS node providing storage for the cloud as a whole?
The Ubuntu cloud docs I read say the minimum number of MAAS nodes is 6. But other OpenStack docs seem to imply that one can run with fewer nodes. What is the minimum number of MAAS nodes do I need to have at least minimal redundancy? And what do I need to do besides installing MAAS nodes, a cluster controller and a region controller and juju? Are there extra steps on top of that to have various levels of HA?
All I'm shooting for at this point is storage redundancy and having VMs (at worst) reboot if the node they had been running on died...
No comments:
Post a Comment