Packet Tracer: BGP Testing

[Verbatim]

Hello,

I've attached a packet tracer file. Will also include router configs. Basically I'm expecting to be able to disconnect one of the paths between the pcs, and have bgp automatically update the routing table and activate the remaining available path. I've only managed to get it to work once, by shutting down port 0/0 on the switch (and I wasn't able to reproduce).

Perhaps the issue is that I'm abusing how these technologies are supposed to work. My goal was to allow link failover between a maximum of 2 devices, one of the 2 links being preferred but less reliable than the other / different speeds (haven't gotten this far, as far as any kind of prioritization). (The different speeds eliminates etherchannel as a solution, but this is really just an exercise in learning / curiousity anyway).

A note on the configs: for some reason, the network line doesn't have the mask, even though I ran the command: network 192.168.x.0 mask 255.255.255.0

PCs are able to ping each other once everything is booted up. Not sure if the configuration of bgp and the vlans is correct; it seems that loopback interfaces are often used rather than vlans in the real world, but when I had bgp setup with loopback interfaces, it wasn't possible for the pcs to directly connect through them; looked like additional routing / bridging would have been required. Maybe that's part of the problem?

 

[Verbatim]

Packet Tracer version 8.2.0.0162

[paul driver]

Hello
Alternatively you can peer the routers via loopback addresses as such if either physical link fails the peering wont, and you will still have reachability to the lan

[MHM Cisco World]

check solution

[Harold Ritter]

Hi @MHM Cisco World ,

It is certainly possible to run multiple parallel bgp sessions between two neighbors. I think the issue here is related to the Packet Trace limitations as @Martin L and @Flavio Miranda mentioned. I would also suggest running this kind of scenario on another simulation platform such as CML or GNS3.

Regards,

[MHM Cisco World]

Can you share example of

bgp multi tcp

session between neighbor.

I want to know how we config

bgp.

Note:- I am talking multi tcp session for one address family not mutli.

[Harold Ritter]

Hi @MHM Cisco World ,

It is as simple as what the original poster configured in his PT file.

You have multiple interfaces between the two peers and you configure a

BGP

neighbor for each interface.

int gi0/0

ip address 192.168.1.1 255.255.255.0

int gi0/1

ip address 192.168.2.1 255.255.255.0

router bgp 65001

neighbor 192.168.1.2 remote-as 65002

neighbor 192.168.2.2 remote-as 65002

address-family ipv4 unicast

neighbor 19.168.1.2 activate

neighbor 192.168.2.2 activate

Regards,

[Martin L]

BGP in PT ?  Bad idea!!!   PT is just a sim and there is no chance that it is behaving like IOS BGP would.  To really learn BGP you would need a real routers or emulator running on IOS or virtual IOS.  I would check Free online Cisco Modeling Lab on Dev Net Sandbox or get GNS3 which requires real router IOS (note only few IOSes are supported).  For quick lab check (routing)  I use GNS3 standalone version with my old 7200 IOS 15.0 image (hardware no longer works but IOS is ok in GNS3).  

Regards, ML
**Please Rate All Helpful Responses **

[Flavio Miranda · Accepted solution]

Hello @Verbatim 

   I took a  look into your file. you did nothing wrong but the problem is the simulator. It will not work the failover as we expect. It works some times but not always.

  You can see that even after you shuting down the path, if you check the command

show ip bgp sum

both neighbors will be UP and it shouldn´t be.  When one path fail the BGP process should put the neighbor as iddle or Active and preffer the other path.

About what you said:

"Perhaps the issue is that I'm abusing how these technologies are supposed to work. My goal was to allow link failover between a maximum of 2 devices, one of the 2 links being preferred but less reliable than the other / different speeds (haven't gotten this far, as far as any kind of prioritization). (The different speeds eliminates etherchannel as a solution, but this is really just an exercise in learning / curiousity anyway)."

 BGP works differently than OSPF or EIGRP or RIP. As PacketTracer only supports EBGP (External BGP), the metric is the PATH which means how many router to the destination. In your scenario both PATH have one hop. You can not use AS prepend on PacketTracer or any other BGP feature like that.

"PCs are able to ping each other once everything is booted up. Not sure if the configuration of bgp and the vlans is correct; it seems that loopback interfaces are often used rather than vlans in the real world, but when I had bgp setup with loopback interfaces, it wasn't possible for the pcs to directly connect through them; looked like additional routing / bridging would have been required. Maybe that's part of the problem?"

It was. The problem with Loopback and eBGP is that you may need to configure

ebgp-multihop&nbsp

; which is not supported on PacketTracer either. On your case, it was not necessary as you stablished neighborship using the Interface and they are directly connected. The layer2 device in between does not count.

  If you had stablished neighborship using the Vlan IP address then it would not work because the physical interface would be considered one hop and you had to use

ebgp-multihop

which is not supported on the simulator.

Botton line, you did everything right but you are using the wrong tool. For basic concept PacketTracer is great but for BGP I suggest GNS3 or Eve.

[MHM Cisco World]

check solution

[Harold Ritter]

Hi @MHM Cisco World ,

> the issue is BGP is slow to detect neighbor down that why the second link not work 

Bear in mind that the default timers are 60sec (keep alive)/180sec (hold time). Have you tried reducing the timers using the

timers bgp 1 3

command?

Regards,

[MHM Cisco World]

the solution with test 

[Verbatim]

Thank you everyone. It does seem Packet Tracer is not the right tool for the job, though have to give it a little credit: it did work once! Haha. I have GNS3 on another laptop, so will try that in the coming days.

Thanks for the suggestion to use IGP rather than BGP. I see BGP a lot at my job, so I went with what I was familiar with. Was planning to implement BFD to speed things up.

[MHM Cisco World]

check solution

[MHM Cisco World]

check solution