Ask Your Question
1

Setting up a new cloud

asked 2014-10-30 08:44:11 -0500

Evanion gravatar image

Hello We are considering migrating our non-virtualized shared hosting environment, over to OpenStack. It feels like we aren't using a lot of the performance that our servers offer, and we want to make better use of them. And at the same time, offer more custom tailored solutions to our more demanding customers.

The server cluster that we want to migrate contains the following hardware:

  1. Internal use & backup manager node
    • 1x AMD 16xCore CPU
    • 32GB RAM
    • 1x 50GB HDD
    • 1x 1TB HDD
    • 1x 500GB HDD
  2. Production node (2x)
    • 2x AMD 16xCore CPU
    • 64GB RAM
    • 2x 500GB HDD
  3. Empty node
    • no CPU (space for 2 AMD)
    • no RAM
    • no HDD (Space for 3)

My plan is to migrate all current customer accounts from one of the production nodes, over to the other. I begin with reprovisioning the internal/backup manager node and stick all the C&C stuff (horizon etc in that.) I can then reprovision the empty production node as a computation node. After that, I spin up a number of instances for the shared environment, and migrate the customers over to the new infrastructure. Last thing I need to do, is to reprovision the second production node as a computation node.

The empty node (node 4) Can either be provisioned as a third computation node, or built for block storage. All nodes are part of a single chassi: http://www.supermicro.nl/aplus/system/2u/2022/as-2022tg-htrf.cfm (http://www.supermicro.nl/aplus/system...)

It might be better to buy a secondary chassi with room for more HDD to build the block storage... we will see.

My question is ... what are the caveats, and how long do you think it would take to set up this cloud (excluding the migration of customers.), and am I planing the infrastructure correctly? or would be be better served doing it differently.

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
1

answered 2014-10-30 12:25:37 -0500

Hi Evanion,

Your on the right way :)

That much (1x AMD 16xCore CPU, 32GB RAM, 1x 50GB HDD, 1x 1TB HDD, 1x 500GB HDD) hardware is not required for Node 1 (Internal use & backup manager node), as it is just a controller node (having Horizon and other components).

I deployed OpenStack many times, based on my experience I can say: Five Hours of time is enough for manual installation and configuration of OpenStack core components (keystone, glance, nova, neutron, cinder, horizon), unless if there are any outstanding issues.

Along with the above specified parameters (RAM, HD, CPUs), please take "Network" (Number of NICs) into consideration.

Controller Needs 3 Nics and each Computation node should have 2 NICs.

edit flag offensive delete link more

Comments

Thanks for your reply. So you think we would be better served buying a lower performance server for the controller, and increase the hardware of node 1 to be on par with node 2 & 3, as a compute node instead?

Evanion gravatar imageEvanion ( 2014-10-30 17:14:38 -0500 )edit

As of now no need to increase the hardware to Node1 to make it as a compute node. Later point if compute resources got exhausted then we can increase the hardware.

TechPatron gravatar imageTechPatron ( 2014-10-30 23:07:14 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower

Stats

Asked: 2014-10-30 08:44:11 -0500

Seen: 504 times

Last updated: Oct 30 '14