Thanks Eugene, now I understand where the library is integrated. But why do you think, batching will not be effective here?
Don't you think https://github.com/openstack/neutron/blob/0363ec23e744a5cae0f85318031e3acaa9e45d15/neutron/agent/ovsdb/impl_idl.py#L79-L92 this will be much faster if the operations are batched in a txn?
Savings here will not be as high as the rootwrap/vsctl case, but vsctl is still the default https://github.com/openstack/neutron/blob/0363ec23e744a5cae0f85318031e3acaa9e45d15/etc/neutron/plugins/openvswitch/ovs_neutron_plugin.ini#L43-L46
I've read twilson's (bp patch author) blog on this http://blog.otherwiseguy.com/replacing-ovs-vsctl-calls-with-native-ovsdb-in-neutron/
I think, it will definitely be faster in both cases if we batch the operations, especially tunnel creations when there are many nodes.
Thanks Eugene, now I understand where the library is integrated.
But why do you think, batching will not be effective here?
Don't you think /github. com/openstack/ neutron/ blob/0363ec23e7 44a5cae0f853180 31e3acaa9e45d15 /neutron/ agent/ovsdb/ impl_idl. py#L79- L92
https:/
this will be much faster if the operations are batched in a txn?
Savings here will not be as high as the rootwrap/vsctl case, but vsctl is still the default /github. com/openstack/ neutron/ blob/0363ec23e7 44a5cae0f853180 31e3acaa9e45d15 /etc/neutron/ plugins/ openvswitch/ ovs_neutron_ plugin. ini#L43- L46
https:/
I've read twilson's (bp patch author) blog on this http:// blog.otherwiseg uy.com/ replacing- ovs-vsctl- calls-with- native- ovsdb-in- neutron/
I think, it will definitely be faster in both cases if we batch the operations, especially tunnel creations when there are many nodes.