cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
6406
Views
5
Helpful
8
Replies

Trunking between 2 3750 switches connected via a wireless bridge

dtran
Level 6
Level 6

Hi all,

I have a Microwave wireless bridge connecting two 3750 switches. The wireless bridge is connected to a layer 3 port on the 3750 at each end and all is working fine today. I have some application requirements that I need to connect the wireless bridge to a layer 2 port, basically a trunk port on the 3750 at each end. Please see attached diagram.

has anyone done this before ? I appreciate any inputs / suggestions.

Thanks

D.

network diagram.JPG

4 Accepted Solutions

Accepted Solutions

Eugene Lau
Cisco Employee
Cisco Employee

G'day,

I've seen this done before where the wireless bridge simply acts as a transport mechanism. If you have a stable microwave link, there's no real issues but obviously this type of set up is more susceptible to interference issues.

Example: Figure 4 outdoor wireless VLANs deployment with Cisco equipment

http://www.cisco.com/en/US/products/hw/wireless/ps430/prod_technical_reference09186a00801444a1.html

Points of interest

- Be aware of any limitations with MTU for your microwave link and ensure it can accomodate any extra bytes if you are using 802.1q trunking

18bytes for ethernet header inc CRC

4 byte for the 802.1q tag

1500bytes for the IP MTU

total = 1522bytes.

Brain storming a little

- unstable trunk links will affect switches that carry the same VLANS as the trunk link - typical Layer 2 spanning-tree type issues

- if you intend to use a redundant links, you'll need to consider the fact that if the wireless goes down, the links on the switch will stay up, potentially black holing traffic, where as Layer 3 routing protocols have hello packets which could detect if the link is down.

You could use probes and scripts such as EEM to automate these kinds of things.

That's about all I can think of right now

HTH

Eugene

View solution in original post

Welcome

In the original Layer 3 topology - where WAN and Microwave are Layer 3 connections - redundancy is quite simple because your subnets are segmented and if anything fails, the routing topology should converge accordingly.

In the new topology - where the WAN is Layer 3 and Microwave is Layer 2 - redundancy considerations become more challenging.

I'll assume that all VLAN's will be bridged across the Microwave, so logically it will look like a big LAN. Let's use subnet 1.1.1.x as an example

---- RouterA ---- SwitchA -- PCA 1.1.1.20

       |L3               |L2

---- RouterB---- SwitchB -- PCB 1.1.1.40

The challenge will be if L2 (microwave) goes down - the 1.1.1.x subnet gets divided. When this happens, PCA cannot talk to PCB.

Internet traffic could potentially have an issue on the return path as well.

So really, it depends on:

- What specifically is driving you to think about changing to Layer 2 and how badly do you need it?

- how many VLAN's across the link (can you have a mix of L2 and L3?) - eg. If you only had 2 devices in one VLAN that needed this topology, it could be easier to manage. If you needed to change the all devices to fit this new topology, it would be much more complex.

- What is the appropriate Disaster Recover Plan that the business can accept? ie. Can you live without redundancy? How big of an outage is acceptable to the affected services

Would need some time to think about any features that could be utilised at the routers such as interface tracking, bridging over T1's etc gets complex but an perfect redundancy solution with the current set of equipment may not be possible when one layer2 network is divided in such a way.

Eugene.

View solution in original post

Hey Danny,

My personal view is that we can always get things to work - a solution, but each solution would have different strengths and weaknesses where some times the weakness is acceptable. I also like to keep things as simple as possible

In the T1/L3 and L2 Microwave topology, if L2 microwave link fails, the L2 domain is split and you can route out of the LAN but there'll be issues routing back in because the default gateway interface on the LAN would be both up and still in the routing topology. Local subnet connectivity is broken.

One way around this is to bridge across the T1's and use BVI's as the default gateway. This essentially turns the L3 T1 bundle into an L2 link. You'd use STP to ensure that forwarding is across the microwave link (which I assume the data is free here ) This way all traffic is through microwave.

Upon failure, STP should make the T1 bundle forwarding so that devices in both buildings can hit the active default gateway interface and be routed accordingly. While bridging typically has performance impacts on the CPU of a router, this would be temporary until the microwave link is rectified.

It would look like

--- routerA ---- switchA

       ||| L2           | L2

--- routerB---- switchB

HSRP for L3 gateway redundancy on BVI should a router fail

T1 bundle in L2 bridge should microwave fail (need to verify whether your  router modules/required features on WAN would support IRB - integrated routing and bridging)

If you had a fiber connection between the buildings then the T1 bundle is not required for building to building connectivity and redundancy. I would make the fiber primary link, microwave secondary through STP port costs.

HTH

Eugene

View solution in original post

Welcome Danny,

Without taking constraints such as cost, scalability, traffic pattern changes, migration requirements into consideration (aka. all the business stuff) - Technically, the Metro link would be very simple solution in creating the required L2 domain as it would act as an extension of your ethernet environment (it just doesn't belong to you )

Typically, SP's and enterprise can employ technologies such as VPLS, q-in-q to provide a transparent connection to the Metro, then you'd simply trunk your link to the SP.

Ultimately, it simply depends on what resources you have at our disposal and the constraints that apply. You can make it work with what you have. You can optimise what you have or you can change it for the future. Maybe if you save the business money, they can put some in your pocket (we can always wish)

HTH.

Eugene

View solution in original post

8 Replies 8

Eugene Lau
Cisco Employee
Cisco Employee

G'day,

I've seen this done before where the wireless bridge simply acts as a transport mechanism. If you have a stable microwave link, there's no real issues but obviously this type of set up is more susceptible to interference issues.

Example: Figure 4 outdoor wireless VLANs deployment with Cisco equipment

http://www.cisco.com/en/US/products/hw/wireless/ps430/prod_technical_reference09186a00801444a1.html

Points of interest

- Be aware of any limitations with MTU for your microwave link and ensure it can accomodate any extra bytes if you are using 802.1q trunking

18bytes for ethernet header inc CRC

4 byte for the 802.1q tag

1500bytes for the IP MTU

total = 1522bytes.

Brain storming a little

- unstable trunk links will affect switches that carry the same VLANS as the trunk link - typical Layer 2 spanning-tree type issues

- if you intend to use a redundant links, you'll need to consider the fact that if the wireless goes down, the links on the switch will stay up, potentially black holing traffic, where as Layer 3 routing protocols have hello packets which could detect if the link is down.

You could use probes and scripts such as EEM to automate these kinds of things.

That's about all I can think of right now

HTH

Eugene

Hi Eugene,

Your feedback is really helpful and I appreciated very much !!!

The microwave wireless link is very stable and I have no issues with it. I actually have a second link between the two buildings, please see the new diagram below and let me know if traffic would failover to the 4 bonded T1's when the wireless link is down. I am not sure what would happen when the wireless link is down in this setup.

Your response is highly appreciated and thanks again Eugene !!!

D.

Welcome

In the original Layer 3 topology - where WAN and Microwave are Layer 3 connections - redundancy is quite simple because your subnets are segmented and if anything fails, the routing topology should converge accordingly.

In the new topology - where the WAN is Layer 3 and Microwave is Layer 2 - redundancy considerations become more challenging.

I'll assume that all VLAN's will be bridged across the Microwave, so logically it will look like a big LAN. Let's use subnet 1.1.1.x as an example

---- RouterA ---- SwitchA -- PCA 1.1.1.20

       |L3               |L2

---- RouterB---- SwitchB -- PCB 1.1.1.40

The challenge will be if L2 (microwave) goes down - the 1.1.1.x subnet gets divided. When this happens, PCA cannot talk to PCB.

Internet traffic could potentially have an issue on the return path as well.

So really, it depends on:

- What specifically is driving you to think about changing to Layer 2 and how badly do you need it?

- how many VLAN's across the link (can you have a mix of L2 and L3?) - eg. If you only had 2 devices in one VLAN that needed this topology, it could be easier to manage. If you needed to change the all devices to fit this new topology, it would be much more complex.

- What is the appropriate Disaster Recover Plan that the business can accept? ie. Can you live without redundancy? How big of an outage is acceptable to the affected services

Would need some time to think about any features that could be utilised at the routers such as interface tracking, bridging over T1's etc gets complex but an perfect redundancy solution with the current set of equipment may not be possible when one layer2 network is divided in such a way.

Eugene.

Hello Eugene, again I very much appreciate your help.

I foresee the complexity and the challanges in accomplishing redundancy with the new topology and redundancy is a requirement (a must have). Your assumption is correct that all VLAN's will be bridged across the Microwave, so logically it will look like a big LAN.

You don't think redundancy can be accomplish with the new topology ? What if I have a fiber link connecting the two 3750's at Layer 2 instead of the wireless microwave link, would I have redundancy if the fiber link fails ?

Thanks Eugene !!!

Danny

Hey Danny,

My personal view is that we can always get things to work - a solution, but each solution would have different strengths and weaknesses where some times the weakness is acceptable. I also like to keep things as simple as possible

In the T1/L3 and L2 Microwave topology, if L2 microwave link fails, the L2 domain is split and you can route out of the LAN but there'll be issues routing back in because the default gateway interface on the LAN would be both up and still in the routing topology. Local subnet connectivity is broken.

One way around this is to bridge across the T1's and use BVI's as the default gateway. This essentially turns the L3 T1 bundle into an L2 link. You'd use STP to ensure that forwarding is across the microwave link (which I assume the data is free here ) This way all traffic is through microwave.

Upon failure, STP should make the T1 bundle forwarding so that devices in both buildings can hit the active default gateway interface and be routed accordingly. While bridging typically has performance impacts on the CPU of a router, this would be temporary until the microwave link is rectified.

It would look like

--- routerA ---- switchA

       ||| L2           | L2

--- routerB---- switchB

HSRP for L3 gateway redundancy on BVI should a router fail

T1 bundle in L2 bridge should microwave fail (need to verify whether your  router modules/required features on WAN would support IRB - integrated routing and bridging)

If you had a fiber connection between the buildings then the T1 bundle is not required for building to building connectivity and redundancy. I would make the fiber primary link, microwave secondary through STP port costs.

HTH

Eugene

Hi Eugene,

Your feedback is extremely helpful and I appreciate it very much !!!

I am still trying to figure out the best path to take on this. Redundancy is a must have for me and it sounds like having T1/L3 and wireless bridge L2 would be too involve and complex to achieve redundancy in this setup.

I still would like to connect the Microwave wireless bridge to a Trunk port on both sides. Now would it make the topology any simpler if I replace the bonded T1's with Metro Ethernet ? Do you know if Metro Ethernet can be connected at L2 (directly to a switch trunk port) ?

Thanks Eugene !!! very much appreciate your help !!!!!!

Danny

Welcome Danny,

Without taking constraints such as cost, scalability, traffic pattern changes, migration requirements into consideration (aka. all the business stuff) - Technically, the Metro link would be very simple solution in creating the required L2 domain as it would act as an extension of your ethernet environment (it just doesn't belong to you )

Typically, SP's and enterprise can employ technologies such as VPLS, q-in-q to provide a transparent connection to the Metro, then you'd simply trunk your link to the SP.

Ultimately, it simply depends on what resources you have at our disposal and the constraints that apply. You can make it work with what you have. You can optimise what you have or you can change it for the future. Maybe if you save the business money, they can put some in your pocket (we can always wish)

HTH.

Eugene

Thanks Eugene !!!!!

I've been looking into Metro Ethernet for a while so this might be a good opportunity for me to push it through management. The 4-bonded T1s can't really support the amount of traffic that's going between the two buildings if the Microwave wireless bridge were to go down. So for full redundancy I would have to bring in a Metro Ethernet link or another Microwave wireless bridge.

So if I were to have a Metro Ethernet link and a Microwave wireless link connecting the two buildings. How would I configure these links for failover in this setup ? Can I configure these links in an Etherchannel ? or Do I configure one link as primary and the other as backup using Spanning Tree port cost ? I like your idea of keeping things simple.

Also, do you have any documentation on VPLS, q-in-q or any other flavors of Metro Ethernet ? I have not work with Metro Ethernet before, this will be a learning experience for me.

Very much appreciate your help Eugene !!!

Danny