Ask Your Question

Revision history [back]

What if the network load on one of the Instances became much higher than other Instances running, which would result in network slowdown for other instances?

Yes it would. As an admin, however, you can use Neutron’s Quality of Service features to limit bandwidth, provide minimum bandwidth and prioritize certain types of traffic.

What if the Processing load (Compute resources) demand increases in any if the running instances that can effect other systems performance?

Instances get a certain number of virtual CPUs and are limited this way. Nova’s filtering scheduler can be tuned to select compute nodes with low CPU load when a new instance is launched. Flavors can also be used to limit CPU cycles and IO.

There are many more things that can happen if look at one's use case more seriously. So I was wondering that is there a service, a functionality or something like that present in Openstack that can help me to counter these probelms. Likewise if there is something where we can limit the resources utilisation of an instance upto some extent.

You could also set up alarms based on metrics like CPU usage and implement your own actions if an instance behaves badly.

I have read something about Octavia but will that service be of any help to counter these problems?

Octavia is a load balancer implementation and doesn’t address the problem of instances using up too many resources.