cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
487
Views
2
Helpful
4
Replies

Slow performance over VXLAN SVI, Nexus

als95
Level 1
Level 1

Hello!

I have a pretty standard XVLAN setup.  In that setup server "A" with IP 10.0.0.100/24 can successfully reach server "B" with IP 10.0.0.200/24. Iperf3 performance if very good - line rate.

My goal is to  have additional IPs on those servers, server "A" assigned IP 203.0.113.1/32 and server "B" assigned IP 203.0.113.2/32

[ srvA 203.0.113.1/32-lo + 10.0.0.100/24-eth0 ] <-> |vxlan-anycast gw 10.0.0.254/24| <-> [srvB 10.0.0.200/24-eth0 + 203.0.113.2/32-lo]

Those /32 IP are BGP propagated

Server's have only `default` route via anycast gateway 10.0.0.254

Output from the Nexus C9336C-FX2 switch. (Both 10.0.0.100 and 10.0.0.200 are on the same switch)

# show bgp vrf myvrf999 ipv4 unicast neighbor 10.0.0.100 routes

*>e203.0.113.1/32 10.0.0.100 0 65000 i

# show bgp vrf myvrf999 ipv4 unicast neighbor 10.0.0.200 routes

*>e203.0.113.2/32 10.0.0.200 0 65000 i

There is a reach-ability between those IPs:

root@serverA:~# tracepath -n 203.0.113.2
1?: [LOCALHOST] pmtu 1500
1: 10.0.0.254 0.682ms
1: 10.0.0.254 0.558ms
2:  203.0.113.2 0.409ms reached
Resume: pmtu 1500 hops 2 back 1

However iperf3 performance is very poor for those IPs

serverA:~# iperf3 -c 203.0.113.2 -p 5001 --bind 203.0.113.1
Connecting to host 203.0.113.2, port 5001
[ 5] local 203.0.113.1 port 38739 connected to 203.0.113.2 port 5001
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 294 KBytes 2.41 Mbits/sec 5 4.24 KBytes
[ 5] 1.00-2.00 sec 0.00 Bytes 0.00 bits/sec 13 11.3 KBytes
[ 5] 2.00-3.00 sec 0.00 Bytes 0.00 bits/sec 1 15.6 KBytes
[ 5] 3.00-4.00 sec 0.00 Bytes 0.00 bits/sec 20 9.90 KBytes
[ 5] 4.00-5.00 sec 0.00 Bytes 0.00 bits/sec 4 7.07 KBytes
^C[ 5] 5.00-5.39 sec 0.00 Bytes 0.00 bits/sec 0 7.07 KBytes

Same test with the 10.0.0.x/24 IPs shows great throughput 

root@serverA:~# iperf3 -c 10.0.0.200 -p 5001 --bind 10.0.0.100
Connecting to host 10.0.0.200, port 5001
[ 5] local 10.0.0.100 port 42939 connected to 10.0.0.200 port 5001
[ ID] Interval Transfer Bitrate Retr Cwnd
[ 5] 0.00-1.00 sec 3.03 GBytes 26.0 Gbits/sec 0 1.84 MBytes
[ 5] 1.00-2.00 sec 2.58 GBytes 22.2 Gbits/sec 0 1.84 MBytes
[ 5] 2.00-3.00 sec 2.33 GBytes 20.0 Gbits/sec 0 3.02 MBytes
^C[ 5] 3.00-3.69 sec 1.57 GBytes 19.5 Gbits/sec 0 3.02 MBytes

What could be the issue there?

 

 

1 Accepted Solution

Accepted Solutions

als95
Level 1
Level 1

Yes, culprit was a re-use of the VNI in the fabric for the unrelated VLAN. 

View solution in original post

4 Replies 4

Dawei
Cisco Employee
Cisco Employee

Do you have enabled "no ip redirect" under your gw SVI? The most likely reason is that for some reason the switch doesn't hardware forwarding these packets.

als95
Level 1
Level 1

Looks like the issue may be related to re-use of the VNI in the fabric. I'm going to fix that and re-check.

als95
Level 1
Level 1

Yes, culprit was a re-use of the VNI in the fabric for the unrelated VLAN. 

Review Cisco Networking for a $25 gift card