LAG with PTP820 1+1 Hot Standby


Can someone help me understand how I can acheive full redundancy using LACP and/or LAG? I have looked for days, every LAG configuration example put out by Cambium so far is not similar to my setup.

Attached is my topology.

How can I configure the cambiums to guarantee operation, no matter if I lose an ODU, an ethernet LINK, or a router?

Hi Cjanik, what you want to setup should be possible either using an optical splitter (for data) or LAG/LACP.  It is going to vary slightly depending on whether it is a PTP820S (1+1 HSB) or PTP820C (2+2 HSB).  Your diagram references a PTP820S.  I've attached a document that shows the physical setup (page 10), and the technical description, installation guide and User guides go into additional detail.  If you use LINKPlanner to build out this link, it is going to assume a single fiber connection (using optical splitter) back into the network.  You can also have two independent fiber connections and use LAG/LACP to do the traffic routing.

The management is going use a splitter and a PROT cable setup.  The splitter cable is shown in this post: and the PROT cable can be made (pin-outs in the technical description) or purchased.  LINKPlanner will default this dual out of band management setup.

I would reach out to your distributor or local Cambium sales team for additional help.

1 Like


Thank you for the documentation! It has helped me understand a couple of things.

I have PTP820S radios.

I do not use space diversity or ACM. The radios are directly connected to a coupler on the dish. The radios are powered by an external power source (not using POE).

TWO things though:

1) All of the documentation mentions using an optical splitter to connect hot/standy radios. This is where I am misunderstanding what I need to do because I am utilizing copper, which has a different physical setup.

2) I plan on utilizing two data lines for traffic - this is for the purpose of having redundancy to the routers. (in case one router drops, the other will take over). I cannot find any documentation regarding how to utilize this setup.

please see attached - this a more detailed overview of what I'm trying to do - and might be doing it wrong?

Here's the scenario of what is happening with this setup:

Eth1 of HOT radio and Eth1 of STBY radio are connected to Router A as a single LAG group with LACP. When HOT radio goes down and STBY becomes active, it takes about 10 seconds before ETH1 of STBY becomes active, when it should be 50ms max. I think it has to do with my radios being configured imporperly - This is where I am asking for assistance. Do I maybe need to change the physical configuration or software configuration?

In addition to that, I will also need to configure Eth2 on HOT and STBY radios to take over for Eth1 HOT and STBY in case Router A fails and Router B has to take over.

Hi, what you are referring to is 1+1 HSB Line Protection Mode (vs. Split Protection Mode with optical splitter).  There is more detail in the PTP820S Technical Description Document starting on page 52 of v10.0.  The main caveat is that line protection mode is only certified on Cisco switches supporting LACP.  Page 140 of the PTP820C&S User Guide v10.0 goes into more detail on the actual configuration.

I'm not sure about a redundant router setup, but I believe, from a PTP820 perspective, it is the same setup as two different network inputs into the radio.  Essentially the PTP820 would have it's ETH1 and ETH2 ports in a LAG configuration with a PIPE service to the radio interface.  The networking equipment would be responsible for deciding which (or both) ETH interfaces are sending data.  Page 123 of the PTP820C&S User Guide v10.0 goes into more detail.

Most important for you would be following the instructions on page 140 of the PTP820C&S User Guide v10.0.  LACP must be enabled on the router ports.  Now, enable LACP on the Ethernet interface connected to the external router (i.e. ETH1) on the active PTP820:

i Select Ethernet > Interfaces > Logical Interfaces. The Logical Interfaces page opens .
ii Select the interface and click Edit. The Logical Interfaces ‚Äď Edit page opens.

iii In the Interface Mode field, select LACP.
iv Click Apply, then Close.

In your case, repeat for ETH2.  Once you copy to mate, the standby unit will have the same configuration.

Also, just for reference, the PTP820 has the ability to do automatic state propagation for controlling port status based on radio status (degredation, up/down, etc.).  It is beyond what you are trying to do, but may be useful in the future.  I've also attached a document that isn't applicable to your current design, but may be also useful in the future.

Ensure that you are running v10.5 of the firmware, and you may need to open a support case if you have the above setup correct and are still seeing significant delay in failover.

1 Like


Thank you for getting back to me.

We are currently on firmware.

Based on your information and reading the documentation you suggested, I have created a topology that I am going to apply and try out. Attached is the topology. Does this look right?

This is a PTP820S 1+1 HSB setup, correct?  Assuming it is, then you can't do the PIPE services as shown in your diagram.  A PIPE service can only have 2 service points (i.e. ETH1 and Radio1), and an interface can only have 1 PIPE service associated with it.  

I would recommend a LAG setup on the PTP820S, with ETH1 and ETH2 as members (and LACP enabled per the other instructions on each ETH interface).  Then the PIPE service will be setup with two service points: the LAG group and Radio Slot 2: Port 1.

Setting up 1+1 HSB through the GUI can be a bit tricky.  Be sure to follow the instructions and do each step in order.  Otherwise you may end up copying the configuration from the 'mate' to the active unit instead, and wipe out your configuration.  You may also get really proficient at factory defaulting these units ;).  Definitely set it up in a lab and verify that the failover works prior to install.  v10.5 of the firmware should work, but v8.3 is guaranteed to work (I still recommend v10.5 and it will force a factory default going backwards).

Side note: very nice diagrams ;)