|Kloxo Enterprise Integration [message #74511]
||Thu, 13 May 2010 13:52
Registered: January 2009
This could go in the Kloxo Enterprise or the HyperVM forums to be sure. |
We are back to re-evaluating LXadmin, errr, Kloxo Enterprise now as the dust settles so we can migrate away from the imploding HSphere hosting platform. A while back I had installed a complete LXadmin cluster but pulled the plug and switched to HSphere as I was concerned about the long-term prospects of LXAdmin. Ironic, but now HSphere is the one that's in trouble. Since I have to go through a platform re-deployment and client migration this is what I want to do.
I'd like to deploy Kloxo Enterprise but each server should live in a VM to facilitate backups and be able to increase capacity more easily. That is, it should be easy to perform a live migration of a VM to another physical node if necessary. It should also be possible to do a bare-metal restore if something bad might happen. A VM-based architecture can make this all happen.
On the one hand, you could deploy Kloxo Enterprise on VMware containers. Or some other commercial offering. But imagine the power if you could do this from within, say, HyperVM!
Or maybe that should be HyperVM integrated with Kloxo?
Let's select the latter to explore this.
You login to Kloxo HyperVM Enterprise (KHE, a fictitious name), you add physical nodes to the configuration. Then when you add a logical server you are given the option to configure it as a traditional node or as a container for one or more VMs. By doing the former you progress with business as usual in the legacy mode. If you do the latter, however, you enable the ability to add services to VMs controlled from within KHE.
Cool so far? Okay!
Now imagine that you want to sell HyperVM VPS. You would be able to manage this as well from within KHE. Instead of configuring a traditional VWS for a client you would instead select VPS, and the VPS would be configured on an available node that's been selected as a VPS resource for client use.
With this architecture we totally create an abstraction between the application(s) and the hardware. IMHO, one should be able to build out the cluster by adding additional nodes at will and then migrating VMs onto new hardware in order to increase capacity. That means, you could start off with a 4 server physical architecture. Dedicate one server to the KHE control panel and then use the remaining three as nodes for hosting 4 logical servers: mail, web, dns, db. As your needs grow, you can add a node and then migrate mail (as an example) to its own machine or perhaps to just a more capable node.
Let me say this right now: AFAIK, NO ONE offers this right now either commercially or as open source.
I'm looking for feedback on this. Furthermore, if we add file storage as an additional resource you would be able to really dumb down the physical nodes further by pushing your disk redundancy into one capable system (NetApp or whatever) instead of trying to deal with RAID and ever-growing storage needs at the individual node level.