Yes, my only concern in using

arunkv1980 · ‎05-05-2014

i have 2 x 20 Mbps Point to Point ethernet links configured with equal ospf cost and both links terminated on the single router at both ends.

interface FastEthernet0/0/0

bandwidth 20000
ip address x.x.x.x y.y.y.y
load-interval 30
duplex auto
speed auto
traffic-shape rate 20000000 500000 500000 1000
!
interface FastEthernet0/0/1
bandwidth 20000
ip address x.x.x.x y.y.y.y
load-interval 30
duplex auto
speed auto
traffic-shape rate 20000000 500000 500000 1000

FastEthernet0/0/0 is up, line protocol is up
Internet Address 135.254.193.53/30, Area 0.0.0.1
Process ID 3435, Router ID 135.254.0.95, Network Type BROADCAST, Cost: 5

FastEthernet0/0/1 is up, line protocol is up
Internet Address 135.254.193.45/30, Area 0.0.0.1
Process ID 3435, Router ID 135.254.0.95, Network Type BROADCAST, Cost: 5

I see majority of the traffic using the link F0/0/0 and only 30% traffic using the second link F0/0/1.

I used per-packet load-sharing on these interfaces which caused throughput problems, so i removed that from the interface. Should i add some CEF commands or should try per-destination load sharing to see equal traffic on both links ?

Jan Rolny · ‎05-05-2014

Hi,

could you please post sh ip route <destination_network> ?

Per-destination load balancing means the router distributes the packets based on the destination address. Given two paths to the same network, all packets for destination1 on that network go over the first path, all packets for destination2 on that network go over the second path, and so on. This preserves packet order, with potential unequal usage of the links. If one host receives the majority of the traffic all packets use one link, which leaves bandwidth on other links unused. A larger number of destination addresses leads to more equally used links. To achieve more equally used links use IOS software to build a route-cache entry for every destination address, instead of every destination network, as is the case when only a single path exists. Therefore traffic for different hosts on the same destination network can use different paths. The downside of this approach is that for core backbone routers carrying traffic for thousands of destination hosts, memory and processing requirements for maintaining the cache become very demanding.

Here is good explanation of OSPF load balancing:

http://www.cisco.com/c/en/us/support/docs/ip/border-gateway-protocol-bgp/5212-46.html

HTH,

Jan

arunkv1980 · ‎05-05-2014

Here is the routing table.

Gateway of last resort is 135.254.193.53 to network 0.0.0.0

10.0.0.0/32 is subnetted, 1 subnets
O E2 10.252.0.189 [110/1] via 135.254.193.53, 1w3d, FastEthernet0/0
[110/1] via 135.254.193.45, 1w3d, FastEthernet0/2/0
135.254.0.0/16 is variably subnetted, 8 subnets, 3 masks
C 135.254.0.89/32 is directly connected, Loopback10
O E2 135.254.0.95/32 [110/20] via 135.254.193.53, 1w3d, FastEthernet0/0
[110/20] via 135.254.193.45, 1w3d, FastEthernet0/2/0
O 135.254.193.192/30 [110/6] via 135.254.193.53, 1w3d, FastEthernet0/0
[110/6] via 135.254.193.45, 1w3d, FastEthernet0/2/0
C 135.254.193.48/30 is directly connected, FastEthernet0/1
C 135.254.193.52/30 is directly connected, FastEthernet0/0
C 135.254.193.44/30 is directly connected, FastEthernet0/2/0
S 135.254.202.0/24 [1/0] via 135.254.193.50
S 135.254.175.0/24 [1/0] via 135.254.193.50
135.250.0.0/32 is subnetted, 1 subnets
O 135.250.0.11 [110/6] via 135.254.193.53, 1w3d, FastEthernet0/0
[110/6] via 135.254.193.45, 1w3d, FastEthernet0/2/0
O*E2 0.0.0.0/0 [110/1] via 135.254.193.53, 1w3d, FastEthernet0/0
[110/1] via 135.254.193.45, 1w3d, FastEthernet0/2/0

in-aricent01-ertr-01#show ip route 0.0.0.0
Routing entry for 0.0.0.0/0, supernet
Known via "ospf 3435", distance 110, metric 1, candidate default path, type extern 2, forward metric 6
Last update from 135.254.193.45 on FastEthernet0/2/0, 1w3d ago
Routing Descriptor Blocks:
* 135.254.193.53, from 135.250.0.11, 1w3d ago, via FastEthernet0/0
Route metric is 1, traffic share count is 1
135.254.193.45, from 135.250.0.11, 1w3d ago, via FastEthernet0/2/0
Route metric is 1, traffic share count is 1

e.ciollaro · ‎05-05-2014

As far as I know CEF per-destionation load-balancing decide which link to use making an hash of source IP and destionation IP. Because per-destination load balancing depends on the statistical distribution of traffic, load sharing becomes more effective as the number of source-destination host pairs increases. (http://www.cisco.com/c/en/us/td/docs/ios-xml/ios/ipswitch_cef/configuration/15-mt/isw-cef-15-mt-book/isw-cef-load-balancing.html#GUID-303F07B8-A52B-4BD3-A1F8-54DF24493D8E)

An alternative algorithm avaiable in CEF is per-port load-balancing which considers also ports in hashing functions. It could be useful if you have just a few sessions (sourceIP - destionationIP). How many session is active on those links ?

E

arunkv1980 · ‎05-05-2014

Hundreds of sessions are in use over the two wan links. So if i make it as per-destination load sharing on the interfaces, then situation will improve that what i have at the moment?

e.ciollaro · ‎05-05-2014

I don't think so because the default load balancing algorithm is "per destination" so probably you are already using it and there are enough sessions to have a good distribution. It's strange it does not balance well. Have these sessions different source/destination IP ?

e

Joseph W. Doherty · ‎05-05-2014

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

As the other posters have already noted, load distribution depends on packet addressing, but it's also impacted by bandwidth demand of individual flows. I.e. If you have a particular flow that's very bandwidth demanding, it could skew the balance even if you have a good distribution of addressing.

If your device supports OER/PfR it can do dynamic load balancing.

arunkv1980 · ‎05-05-2014

I know which source and destination pair consumes more bandwidth and overloading my first wan link. I am thinking of using Policy Based Routing with failover to move this traffic to the second link which is not used much. I am not sure how far it will work.

Let me know if there is better way to do this.

Joseph W. Doherty · ‎05-05-2014

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Sure, PBR make work fine if you can manually balance out "known" bandwidth hogs. However, it can be a PIA to maintain and link failure can sometime be tricky to deal with.

The better way was already noted in my first post, OER/PfR (if supported), as it supports dynamic load balancing. I.e. it can constantly juggle flows to balance your links. It also supports links of different bandwidths.

arunkv1980 · ‎05-05-2014

Yes, my only concern in using PBR is how the traffic will failover when the specific wan link fails.

I mention my router model and IOS version, let me know if it supports OER/PFR. also provide me url to know more about OER/PFR.

Router at Site-A

3800 Software (C3845-IPBASEK9-M), Version 12.4(22)T2, RELEASE SOFTWARE (fc4)

Router at Site-B

2800 Software (C2800NM-ADVIPSERVICESK9-M), Version 12.4(15)T9, RELEASE SOFTWARE (fc5)

Joseph W. Doherty · ‎05-05-2014

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

12.4T release, I recall, supports PfR.

I've run OER/PfR on 3800s and 2800s, but you might need more than IP Base for the feature. [edit - yup, the data sheet found in the next reference notes the features sets required for PfR - should be in your Advanced IP Services version, on your 2800.)

You might start here: http://www.cisco.com/c/en/us/products/ios-nx-os-software/performance-routing-pfr/index.html

Much (much) more if you search Cisco's main web site. Documentation should also be found in your IOS version command reference and guide.

OSPF equal cost load balancing is sending unequal traffic on the links.