[ovs-discuss] About QoS configuration for OpenvSwitch

Xiao Ma (xima2) xima2 at cisco.com
Mon Feb 22 09:52:46 UTC 2016


Hi,

Could anybody tell me why the bandwidth is so unstable?

Thanks!



在 2016年2月15日,下午8:18,Xiao Ma (xima2) <xima2 at cisco.com<mailto:xima2 at cisco.com>> 写道:

Hi, All

I want to use the QoS feature of openvswitch for networks in openstack cloud environment ( neutron in Vlan Mode).

I created four virtual machines, test-left1/test-left2 in vlan 1003 and test-right1/test-right2 in vlan 1004.
test-left1/right-right1 in host2 and test-left2/test-right2 in host3.


And I can see the port created in ovs bridge in host2:
==========================================
[root at host2 home]# ovs-vsctl show
5844eda6-56d4-4711-9697-6eef12a53265
    Bridge br-int
        fail_mode: secure
        Port "qvo583906a4-bb"
            tag: 1004
            Interface "qvo583906a4-bb"
        Port br-int
            Interface br-int
                type: internal
        Port int-br-inst
            Interface int-br-inst
                type: patch
                options: {peer=phy-br-inst}
        Port "qvo6d7a3443-4d"
            tag: 1003
            Interface "qvo6d7a3443-4d"
    Bridge br-inst
        Port phy-br-inst
            Interface phy-br-inst
                type: patch
                options: {peer=int-br-inst}
        Port br-inst
            Interface br-inst
                type: internal
        Port "ens1f1"
            Interface "ens1f1"
    ovs_version: "2.4.0"
==========================================

I tested the bandwidth between left1 and left2 and it’s about 1000M as the physical nic provided.

Then I created the qos queue using the command bellow:

ovs-vsctl set port ens1f1 qos=@newqos -- --id=@newqos create qos type=linux-htb other-config:max-rate=1000000000 \
           queues:1=@queue1 \
           queues:2=@queue2 -- \
       --id=@queue1 create queue other-config:min-rate=200000000 other-config:max-rate=1000000000 -- \
       --id=@queue2 create queue other-config:min-rate=800000000 other-config:max-rate=1000000000


[root at host2 home]# ovs-vsctl list Queue
_uuid               : 98226a7c-270e-4213-a00f-301f464d8222
dscp                : []
external_ids        : {}
other_config        : {max-rate="1000000000", min-rate="200000000"}

_uuid               : a427ea28-704a-4c5b-8e13-27005d31b0e6
dscp                : []
external_ids        : {}
other_config        : {max-rate="1000000000", min-rate="800000000"}

_uuid               : fee9fb4c-f079-477d-a1dc-c6f717eb29e0
dscp                : []
external_ids        : {}
other_config        : {max-rate="1000000000", min-rate="200000000"}

_uuid               : bcaf0695-3796-4639-ae03-07ccb0d6159d
dscp                : []
external_ids        : {}
other_config        : {max-rate="1000000000", min-rate="800000000"}



The problem is after I created the  qos queues, the bandwidth between left1 and left2 became unstable as bellow:
Could anybody explain this to me and help me to solve this problem ?

Thanks.


<未知.png>





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20160222/db74d3d0/attachment-0002.html>


More information about the discuss mailing list