Nexus 7010 OSPF Equal Cost Paths

Hello,
I currently have two physical links connecting one data center to another. These are both 10Gb links and I have manually set the cost to the primary link to '1' and the secondary link to '10'. My question is, if I set the secondary link to '1' they would have equal cost routes. What is the selection process at this point? Will equal cost load balancing automatically kick in and use both links?
Thank you,
Scott

Scott
Haven't used Nexus switches but generally yes it should do depending on the routing protocol ie. statics, EIGRP, OSPF etc. will use equal cost paths if they are in the routing table.
BGP is different in that without further configuration it picks just one path so there is only one entry in the routing table.
By default it will use per destination load sharing and the default on Nexus is destination IP address and port number to choose which link to use.
"sh ip load-sharing"
will show you the current method it is using.
Jon

Similar Messages

OSPF Equal Cost Path Selection

This is a nerdy enough qeury in reality.
We have a single area - area 0.0.0.32. All intra-area routes. We have 2 switches in the core of the network, and 10 switches at the edge. All of these switches are connected via layer 3 OSPF routed links.
The cost for all links is 20 - which is based on bandwidth between the boxes - which is 2Gbps.
Have a look at the enclosed jpeg to get an idea.
Very simple.
Query revolves around the path selection available to OSPF.
The path from Core 2 to the 10.32.51.0 network is easy - straight across the link between the core switches for a cost of 20.
Question :- if the link between the cores fail, which path will be chosen by OSPF and why?
It can go through ANY of the other edge switches for a total cost of 40, but it will choose a particular one. What criteria does OSPF use to select this path?
Remember, the path costs are equal, they are all intra-area.
I tried messing with Router ID, but this doesn't seem to be it.
I tried highest interface IP addresses, but it doesn't seem to be this.
It is not a random act, the algorithm chooses the same one every time.
There must be some parameter in the LSDB that is the defining one when it comes to path selection.
Can you help me out please. I need someone who understands the OSPF algorithm better than I do (which might not be hard!).
Appreciate any comments for debate.

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
My guess (as I haven't re-read the RFC), selection of an ECMP to retain in a routing table (assuming all possible ECMP are not retained) and/or exactly how packets or flows are ECMP routed is implementation dependent.
I recall years ago bumping into a situation where I had 6 (OSPF) ECMP on a Cisco router which had the (then) default allowance of 4 ECMP in the route table. I don't recall exactly what the issue was, but whatever it was doing I considered it a bug. The "fix" was to allow the router to use all 6 ECMP. (Again, don't remember the specifics, but the issue I saw was more involved than 2 of the 6 ECMPs weren't retained.)

OSPF equal cost path

Hi all,
I have
The network 192.168.0.0 is directly connected to Router A and B. Router A and B are redistributing the network in the ospf area 0.
Router A,B,C,D belong to the same OSPF Area.
Router C is DR. The cost of the link is indicated in the draw.
Question:
Router D (and router C) is using as next hop the Router A for reaching the network 192.168.0.0.
Why the router D is not using the router B as next hop due to the fact that all the possible paths are equal cost?
Thanks.

Hi,
At the end the information provided by Rolf fit in my case.
Playing with the cost to reach the ASBR and the metric of the external route I was able to figure out why the router A is the50
preferred exit point: during the redistribution router A is modifying the metric value, in this way C and D are using A as
Thanks all!!!

IS-IS and IPv6 in equal cost path topology

Network of 7609s running 12.2(18)SXE3. Two
equal cost path links between each 7609. IS-IS Level 2 routers only. When pinging an off campus IPv6 host every other packet is lost. It seems like the equal cost paths prevent an alternative path off campus but that path isn't working.
We cannot figure out where to look next.
Any thoughts ? Is this possibly an IS-IS
issue ?
Thanks

OK. I'll try. We have 5-7609s running 12.2(18)SXE3.
Each 7609 has a variety of 100/1000Base Ethernet
interfaces. They each also have one WS-X6704-10GE out
of which connections to our primary and secondary
backbones originate. This is how the routers are
interconnected.
Three of the five core routers are currently running
IS-IS. If we ssh into any router and ping
www.netbsd.org every other packet is lost.
At the end of the network are two 6506 switches each
with a 10GE module and a 1GE module.
Attached to the 1GE module are fiber links two
two different Juniper M10s which also run IS-IS
and through which we get upstream connectivity
to our provider.
All of our routers are level 2 routers in the same area.
Does this help? If not please let me know what else
you would need.
Thanks,Steve

EIGRP- Equal Cost Paths Selection

Hello,
This might be a simple question; using EIGRP and having 2 equal cost paths to the destination; how does the router/L3 switch route select the path from those two equal cost paths? Is there a way to force one path over anthoer?
Thanks in advance.
Best, ~sK

To answer your question we need to be clear that there are 2 parts to the decision about which path will be used to forward traffic. The first part is the identification of paths and their insertion into the routing table. This is the part where EIGRP plays a role. If EIGRP determines that there are two equal cost viable paths then both of the paths will be put into the routing table. This ends EIGRP involvement in the process. The part of the process that takes a particular packet and determines which path to use is handled by CEF. By default CEF looks at the number of available paths toward a destination and when there are multiple paths CEF will use a process that calculates based on source and destination address to choose a particular path. For the same source address and same destination address CEF should choose the same path every time.
HTH
Rick

OSPF Equal Cost (multiple links) but unbalanced traffic

Hi!
I would like to ask about OSPF.
We are using 4 links running OSPF point to point with equal cost to load balance MPLS L2 traffic between our two routers.
And from what we are experiencing right now is that it didnt balance the load the way it should have.
The first link always catches up the bulk of the traffic, while the remaining 3 links load balances, I would like to inquire on steps to mitigate the issue.
Appreciate if I could receive some help to solve the issue
Thank you very much in advance!
Best Regards,
asakurahao

I do have almost the same problem.
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 ?

OSPF equal cost - loops

Hi
Please see attached diagram.
When I traceroute from the Cisco 892 router (on the top) to an 10.31.48.0 address the packages is looped between the two cisco 2821 routers two times before the package is delivered.
Type escape sequence to abort.
Tracing the route to 10.31.48.2
VRF info: (vrf in name/id, vrf out name/id)
1 172.20.1.2 0 msec
    172.20.1.1 0 msec
    172.20.1.2 0 msec
2 10.31.45.2 12 msec
    10.31.45.7 4 msec
    10.31.45.2 4 msec
The package should go from the 172.20.1.2 directly to 10.31.45.2 or 10.31.45.7 and then to the end host. The traceroute shows that the package is bounced two times.
Any suggestions?
Best Regards, Steffen.

If it was a loop, your ping would not work. The output you get is as a result of the probes used in a traceroute. This link seems to give a reasonable explanation:http://www.dasblinkenlichten.com/why-does-my-traceroute-look-like-that/
Just so you can see, run an extended ping on the 892 like below and share output here.
Cisco-892#ping
Protocol [ip]:
Target IP address: 10.31.48.1
Repeat count [5]:
Datagram size [100]:
Timeout in seconds [2]:
Extended commands [n]: y
Source address or interface:
Type of service [0]:
Set DF bit in IP header? [no]:
Validate reply data? [no]:
Data pattern [0xABCD]:
Loose, Strict, Record, Timestamp, Verbose[none]: r
Number of hops [ 9 ]: 3
Loose, Strict, Record, Timestamp, Verbose[RV]:
Sweep range of sizes [n]:

MPLS and dual equal-cost paths

If a site has two outbound WAN routers (for redundancy) that are configured as PE routers with the WAN interfaces configured with MPLS, will two LSPs be created using LDP (one from each PE) allowing the WAN routers to loadbalance traffic across the MPLS backbone?

Load balancing is very much possible in MPLS environment. Following are the general rules for load balancing labeled packets
1. If the MPLS payload is an IPV4 packet, the load balancing is done by hashing the source and destination IP address of the IPv4 header
2.IF the MPLS payload is an IPv6 packet, the load balancing is done by hashing the source and destination IP address of the IPv6 header.
3.If the MPLS payload is not an IPv4 packet, the load balancing is done by looking at the value of the bottom label
Label stack does not have protocol identifier field so to know what is the MPLS payload is, router can look first nibble following the MPLS label stack, if it is 4 then router considers this an IPv4 packet and performs IPv4 CEF hashing. if it is 6 , MPLS payload is considered IPv6, and the router performs IPv6 CEF hashing
For more information you can refer book "MPLS fundamentals" by Luc De Ghein.

Load Balancing with OSPF and maximum-paths command

Hello,
Just a quick query really, we have a disribution layer 3 switch, in its routing table it has 3 default routes all with the same metric from the core router, this is because the core router is setup with the comamnd "default-information originate always metric 50" which obviously proagates the default route around the area and the metric never changes from 50.
So i have a routing table that looks like this:
O*E2 0.0.0.0/0 [110/50] via 77.95.176.9, 06:44:51, GigabitEthernet4/9
[110/50] via 77.95.176.17, 06:44:51, Vlan903
[110/50] via 91.203.72.5, 06:44:51, Vlan262
Three default routes with the same metric, does this mean that the router IOS will load balance traffic over all three routes evenly? I mean i have been reading up on it and appartemtly i dont have the command "maximum-paths 3" under my ospf process?
I have been doing some traceroutes from this switch to the internet (various sites) and all the traffic seems to be going out over the first route in the table that next hop is 77.95.176.9
My question is how can i verify that load balancing is taking place, or if its not then i need to add this "maximum-paths 3" command to the ospf on the local switch? I would say load balancing is not taking place but im sure i have seen traffic from one customer being routes over all 3 paths due to matching spikes on the SNMP sensors?
Many Thanks.
Matt

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
Yes, your traffic should use all three paths, as Rick notes, OSPF, on Cisco, normally defaults to using up to 4 equal cost paths.
As Rick also notes mentioning CEF, how actual traffic is forwarded across ECMP can vary. Often, the device will keep all traffic for the same flow on the same egress port, and attributes selected for actual egress port selection might be deterministic. I.e. it's possible same traffic flow will always be sent to the same egress port. (This means even with ECMP, you may not see an equal load distribution.)

ECMP - equal cost multi path

Hi,
What is the concept behind ECMP (equal cost multi path) ? Is it different for EIGRP , OSPF , ISIS etc ?
thanks

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
The concept behind ECMP is to actively and concurrently take advantage of multiple link/path bandwidth.
Oh, and to just add to what Jon has already noted, ECMP usually doesn't track actual load. So, "seeing" unequal path/link loading, especially short term, isn't unusual.
PS:
There's also unequal cost multi-path routing too - EIGRP supports that.

Nexus 7010 upgrade path

We currently have two Nexus 7010 with 5.0(2a) as system images.
We would need to know the correct upgrade path to 6.1(1). On the release notes it reads the path is from 4.2(8), 5.0(5) or 5.1(6) to 5.2(5) then to 6.1(1).
Also if ISSU is possible or, because we may need to upgrade EPLD, if there is no upgrade path to do a non-disruptive upgrade.

You probably need to dig a little deeper to get a definitive answer (sup1 or 2, type of cards, etc..) but here is a diagram in the release notes for 6.1 found here:
http://www.cisco.com/en/US/docs/switches/datacenter/sw/6_x/nx-os/release/notes/61_nx-os_release_note.html
If this posts answers your question or is helpful, please consider rating it and/or marking as answered.

If equal cost routes exist, OSPF uses CEF load balancing?

Hi All,
Can anyone explain about:
. If equal cost routes exist, OSPF uses CEF load balancing?

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
Rick is correct, but if his response, with mine, causes any confusion. . .
To OP's original question:
If equal cost routes exist, OSPF uses CEF load balancing?
The answer is technically no, for the reason Rick describes.
But if we rephrase, such as:
Does CEF load balance across multiple equal cost routes generated by OSPF?
The answer would be yes.
I suspect the latter question is what the OP really had in mind, but again, Rick is correct to distinguish that OSPF doesn't use CEF.

Nexus 7010 - redistribute host routes in to OSPF

In my Nexus 7010 infrastructure, I have a requirement to redistribute host routes in to OSPF so that the other connected sites receive the host routes through OSPF.
These hosts are part of the connected network (192.168.100.0/24) on Nexus, I dont want this network to be advertised in to OSPF but I want the hosts inside this subnet (192.168.100.11 and 192.168.100.12) to be advertised.
I tried to create a prefix list with these two hosts and added them to a route-map and then using the "redistribute direct" command called this route-map in to OSPF, but this is not working.
However, if I add the entire subnet (the connected network) in to this route-map, it shows up in OSPF.
Is there a way I could have only the hosts to be part of the OSPF database?

Hello, There is no need for VDC or VRF I think. are we missing a trick here. I also just tested for my own sanity. (Or am I completely under-thinking this)
N7K-1 has vlan 24 and the host route will be to 172.25.24.20
N7K-1 Config:
conf t
ip route 172.25.24.20/32 Vlan24
ip prefix-list TEST seq 5 permit 172.25.24.20/32
route-map STATIC_TO_OSPF permit 10
match ip address prefix-list TEST
router ospf 1
bfd
router-id 192.168.101.1
redistribute static route-map STATIC_TO_OSPF
here is what I see on its neighbour N7K2:
DC1-N7K2# show ip route 172.25.24.20
IP Route Table for VRF "default"
'*' denotes best ucast next-hop
'**' denotes best mcast next-hop
'[x/y]' denotes [preference/metric]
'%<string>' in via output denotes VRF <string>
172.25.24.20/32, ubest/mbest: 1/0
    *via 192.168.102.5, Eth8/1, [110/20], 00:00:07, ospf-1, type-2
DC1-N7K2#
                Type-5 AS External Link States
Link ID         ADV Router      Age        Seq#       Checksum Tag
172.25.24.20    192.168.101.1   479        0x80000002 0x0eed    0
So in essence point the host routes statically to the directly connected interface.
hth.
Bilal

Disruptive ISSU 6.1.4a- 6.2.8 on Nexus 7010 sup1 because of LACP timers.

Hi all.
The problem.
Today I updated my Nexus 7010 sup1 from 6.1.4a to 6.2.8.
I want did it in ISSU mode, but after impact check I got this:
Compatibility check is done:
Module bootable Impact Install-type Reason
1 yes non-disruptive rolling
2 yes non-disruptive rolling
3 yes non-disruptive rolling
4 yes non-disruptive rolling
5 yes disruptive reset Some LACP ports not in steady state or operating in 'rate fast' mode.
6 yes disruptive reset Some LACP ports not in steady state or operating in 'rate fast' mode.
7 yes non-disruptive rolling
8 yes non-disruptive rolling
9 yes non-disruptive rolling
10 yes non-disruptive rolling
Additional info for this installation:
Service "lacp" in vdc 1: LACP: Upgrade will be disruptive as 6 switch ports and 0 fex ports are not upgrade ready!!
Issue the "show lacp issu-impact" cli for more details.
(modified the impact to <Hitful> for module <6>)
Do you want to continue with the installation (y/n)? [n] y
I went on with yes and update script reboot both sups after updated all modules.
It was quite a surprise for me (yes I know I must see word "disruptive" opposite my sups 5 and 6). Because I already had done two ISSU updates on two nexuses (from 5.1.* ->5.2.7 and 5.2.7 -> 6.1.4a) and didn`t have any trouble with LACP timers. Is it a new feature of the 6.* train?
I have another Nexus that I want to update. And it also has same problem with LACP timers.
show install all impact give me the same disruptive result because of LACP.
Can I somehow suppress such ISSU behavior? In case of LACP. I don`t have vPC, just ordinal PC.
It is a way better if some LACP interfaces flap in process, than an almost 14 minutes of all 7010 chassis reboot that I had.
Although problem with LACP timers is that they must be the same on the switch side and on the other side. And in case of switches, linux boxes or HP VCs changing LACP timers isn`t a big problem. IT is a biggg problem in case of the Windows Server.
sh lacp interface ethernet 8/13
Interface Ethernet8/13 is up
Channel group is 13 port channel is Po13
Local Port: Eth8/13 MAC Address= 40-55-39-23-1e-c1
System Identifier=0x8000, Port Identifier=0x8000,0x80d
Operational key=12
LACP_Activity=active
LACP_Timeout=Long Timeout (30s)
Neighbor: 0x1
MAC Address= ac-16-2d-a4-f2-54
System Identifier=0xffff, Port Identifier=0xff,0x1
Operational key=17
LACP_Activity=active
LACP_Timeout=short Timeout (1s)
They must be the same and equal 30s for successful ISSU

You probably need to dig a little deeper to get a definitive answer (sup1 or 2, type of cards, etc..) but here is a diagram in the release notes for 6.1 found here:
http://www.cisco.com/en/US/docs/switches/datacenter/sw/6_x/nx-os/release/notes/61_nx-os_release_note.html
If this posts answers your question or is helpful, please consider rating it and/or marking as answered.

Nexus 7010 mgmt0 useage opinion

As a Senior Network Engineer I have entered into a bit of a debate with our Architect about the use of the mgmt0 interfaces on the nexus 7010 switch (dual-sups, M2 and F2 linecards).
I would like to know opinion of the Cisco support network.
I believe the mgmt0 interface should left alone for control plane traffic only and Out Of Band management access (ie ssh). At the moment I have made a subnet for all VDCs with the mgmt0 (vrf management) sitting in a common subnet. The physical mgmt0 interfaces from both SUPs are connected a management hand off switch. The mgmt0s also serves as our control plane for VPCs. The VPC peer-link however is using main interfaces of the line-cards.
The opinions;
- The Architect thinks we should use all the mgmt0 interfaces for snmp, ntp, tacacs netflow-analysis and switch management.
- However, I think I should use a traditional Loopback to perform these functions within the linecards. The mgmt0 should only be used if traditional restricted switch access has failed.
My Basis;
the Loopback never goes down, uses multiple paths (the OOB hand off switch could fail closing switch management access completely). The mgmt0 should be used as a last resort of management access to CMP.
Thoughts please - Cheers

I see your point about wanting to mitigate the impact of losing the OOB switch. I don't think the mgmt0 interface going down is considered the level of failure that will trigger a Supervisor switchover though. That's the way I read the Nexus 7000 HA whitepaper (and what I've seen based on some limited experience with taking apart a 7k pair).
So, no the 7k can't send you an SNMP trap or syslog message if it's configured management path is offline. Mitigation of that could be via your NMS polling the devices's mgmt0 addresses. No response = trouble in paradise. Investigation step would be to log into the 7ks using the loopback IP and local authentication since your TACACS source-interface (mgmt0) is offline and going from there.
The handful I've built (mostly 5k setups) I go for a Cat 3k switch with dual power supplies as the OOB switch. Once one of those is setup and seen not to be DOA, it's generally going to stay up until someone goes in and uplugs it or initiates a system reload.

Nexus 7010 OSPF Equal Cost Paths

Similar Messages

Maybe you are looking for