Problem:
A typical Aruba SDWAN Branch Gateway will have one or more uplinks.
There are times when the uplink being used has combination of Static, DHCP or PPPoE.
Despite configuring all the uplink with same weight say "10", we will see some link taking higher load than the other.
Diagnostics:This problem will occur not due to the weight in Load Balancing but with route weight.
For example:
In the below output we see the load balancing not happening properly. VLAN 4094 has utilized the least bandwidth despite having same weight "10"
#show uplink debug
Uplink Manager: Enabled
Uplink Health-check: Enabled FQDN: pqm.arubanetworks.com(13.234.96.243)
Uplink Load-balancing:Enabled Mode: Round-robin
Uplink Management Table
-----------------------
Uplink Type Properties Uplink-id State Gateway Reachability WAN Type Speed Weight B/w utiln Max b/w
----------- ---------- --------- ----- ------- ------------ -------- ----- ------ --------- -------
Wired vlan 3000 PPPoE Connected 46.93.112.1 Reachable Internet 1.000 Gbps 10 0.01% 100%
Wired vlan 3001 Static1 Connected 192.168.1.1 Reachable Internet 1.000 Gbps 10 2.32% 100%
Wired vlan 3002 Static2 Connected 192.168.2.1 Reachable Internet 1.000 Gbps 10 2.19% 100%
#show datapath uplink verbose
Datapath Uplink Table Entries
-----------------------------
Vlan Srvc-type Link-id Reachable Weight (rweight) Bw-utiln Sessions WAN Excep Sessions WAN hits HA peer tnl Rwgt-curr Max-bw Bw-cap
---- --------- ------- --------- ---------------- -------- -------- ------------------ -------- ----------- --------- ------ ------
3000 Internet 101 YES 10 ( 333) 0.01% 2411 ( 6%) 0 0 0 0 100% 1000 Mbps
3001 Internet 102 YES 10 ( 333) 2.32% 16967 ( 46%) 0 0 0 0 100% 1000 Mbps
3002 Internet 103 YES 10 ( 333) 2.19% 17015 ( 46%) 0 0 0 0 100% 1000 Mbps
Please note the static route for PPPoE is not listing below
show ip route
Codes: C - Connected, O - OSPF, IA - OSPF Inter Area, E1 - OSPF External Type 1
E2 - OSPF External Type 2, B I - BGP Interior, B E - BGP Exterior, S - Static
U - BGW Peer Uplink, M - Management, Ru - Route Usable, * - Candidate Default
V - RAPNG VPN/Branch, I - Crypto-Cfgset, N - Not Redistributed, Bc - Cloud Overlay Protocol
S* 0.0.0.0/0 [50/1] via 192.168.1.1
[50/1] via 192.168.2.1
SolutionThe problem here is that the default Route cost of Static default gateway is 1 where as the Static route cost of PPPoE/DHCP link is 10.
The solution to this problem would be to change the cost of the static default gateway of VLAN 3001 and 3002 from 1 to 10 from Aruba Central.
After the change is done, it is commended to shut the PPPoE link once and unshut it so that the link can participate in load balancing
In the below output we see the load balancing working fine with different uplinks
#show uplink debug
Uplink Manager: Enabled
Uplink Health-check: Enabled FQDN: pqm.arubanetworks.com(13.234.96.243)
Uplink Load-balancing:Enabled Mode: Round-robin
Uplink Management Table
-----------------------
Uplink Type Properties Uplink-id State Gateway Reachability WAN Type Speed Weight B/w utiln Max b/w
----------- ---------- --------- ----- ------- ------------ -------- ----- ------ --------- -------
Wired vlan 3000 PPPoE Connected 46.93.112.1 Reachable Internet 1.000 Gbps 10 1.04% 100%
Wired vlan 3001 Static1 Connected 192.168.1.1 Reachable Internet 1.000 Gbps 10 1.19% 100%
Wired vlan 3002 Static2 Connected 192.168.2.1 Reachable Internet 1.000 Gbps 10 1.13% 100%
#show datapath uplink verbose
Datapath Uplink Table Entries
-----------------------------
Vlan Srvc-type Link-id Reachable Weight (rweight) Bw-utiln Sessions WAN Excep Sessions WAN hits HA peer tnl Rwgt-curr Max-bw Bw-cap
---- --------- ------- --------- ---------------- -------- -------- ------------------ -------- ----------- --------- ------ ------
3000 Internet 101 YES 10 ( 333) 0.01% 10792 ( 33%) 0 0 0 0 100% 1000 Mbps
3001 Internet 102 YES 10 ( 333) 2.32% 10570 ( 32%) 0 0 0 0 100% 1000 Mbps
3002 Internet 103 YES 10 ( 333) 2.19% 10768 ( 33%) 0 0 0 0 100% 1000 Mbps
You can verify the routing table with the updated route of the PPPoE link as well as highlighted below
show ip route
Codes: C - Connected, O - OSPF, IA - OSPF Inter Area, E1 - OSPF External Type 1
E2 - OSPF External Type 2, B I - BGP Interior, B E - BGP Exterior, S - Static
U - BGW Peer Uplink, M - Management, Ru - Route Usable, * - Candidate Default
V - RAPNG VPN/Branch, I - Crypto-Cfgset, N - Not Redistributed, Bc - Cloud Overlay Protocol
S* 0.0.0.0/0 [50/1] via 83.24.145.1
[50/1] via 192.168.248.1
[50/1] via 192.168.249.1