Ask Your Question
0

Network traffic with multiple compute nodes

asked 2011-10-20 09:36:43 -0500

av-sudhir gravatar image

Hi All,

I guess when we have all in one openstack controller with more than one compute nodes with flat dhcp as the networking all the network traffic in & out of the compute nodes into the instances happens through the nova-network interface(192.168.3.1).

When we have hundreds of instances running then the NIC on the nova controller becomes a bottleneck, is there a way instead of passing through the controller all of the traffic to instances happens directly from compute nodes.

C:\test1>tracert 10.144.201.215
Tracing route to http://in01filer04.internal.xyz.com [10.144.201.215]
over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.3.1
2 <1 ms <1 ms <1 ms http://indc-c6500-4.internal.xyz.com [10.144.199.254]
3 <1 ms <1 ms <1 ms http://indc-c6500-1-int1-2.internal.xyz.com [10.144.193.5]
4 <1 ms <1 ms <1 ms http://indc-c6500-5-int6-1.internal.xyz.com [10.144.193.10]
5 <1 ms <1 ms <1 ms http://in01filer04.internal.xyz.com [10.144.201.215]
Trace complete.

Thanks, Sudhir

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2011-10-20 15:05:53 -0500

everett-toews gravatar image

For this you will need the new high availability option in nova-network in Diablo.

http://docs.openstack.org/diablo/openstack-compute/admin/content/existing-ha-networking-options.html (http://docs.openstack.org/diablo/open...) http://unchainyourbrain.com/openstack/13-networking-in-nova (http://unchainyourbrain.com/openstack...)

We haven't implemented this ourselves so I won't be able to help you with any of the details but hopefully that will get you started.

Everett

On Thu, Oct 20, 2011 at 3:40 AM, Sudhir < question175406@answers.launchpad.net > wrote:

New question #175406 on OpenStack Compute (nova): https://answers.launchpad.net/nova/+question/175406 (https://answers.launchpad.net/nova/+q...)

Hi All,

I guess when we have all in one openstack controller with more than one compute nodes with flat dhcp as the networking all the network traffic in & out of the compute nodes into the instances happens through the nova-network interface(192.168.3.1).

When we have hundreds of instances running then the NIC on the nova controller becomes a bottleneck, is there a way instead of passing through the controller all of the traffic to instances happens directly from compute nodes.

C:\test1>tracert 10.144.201.215 Tracing route to http://in01filer04.internal.xyz.com [10.144.201.215] over a maximum of 30 hops:

1 <1 ms <1 ms <1 ms 192.168.3.1 2 <1 ms <1 ms <1 ms indc-c6500-4.internal.xyz.com[10.144.199.254] 3 <1 ms <1 ms <1 ms http://indc-c6500-1-int1-2.internal.xyz.com [10.144.193.5] 4 <1 ms <1 ms <1 ms http://indc-c6500-5-int6-1.internal.xyz.com [10.144.193.10] 5 <1 ms <1 ms <1 ms http://in01filer04.internal.xyz.com [10.144.201.215] Trace complete.

Thanks, Sudhir


You received this question notification because you are an answer contact for OpenStack Compute (nova).

edit flag offensive delete link more

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: 2011-10-20 09:36:43 -0500

Seen: 33 times

Last updated: Oct 20 '11