DX Unified Infrastructure Management

 View Only
  • 1.  Benefits of Tunnel Hub Servers in large environment

    Posted Jun 28, 2022 08:52 AM
    Hi Team,

    We have a large customer with around 10+ K servers being monitored. The architecture has Primary & HA hubs, Tunnel Hub Servers and Remote Hubs. The robots are reporting to Remote hubs. Remote hubs have tunnels / queues created to Tunnel Hub servers. Tunnel hub servers have tunnel / queues created to primary / HA hubs.

    My Question is: in such a environment, what are the benefits of Tunnel Hub servers?

    My understanding is, it's just being used as intermediary only to receive alerts / qos from remote hubs and pass them to primary / HA hubs. The setup will work without Tunnel Hub servers by creating queues directly between remote and primary / HA hubs.

    Please suggest.


    Jitendra Sharma

  • 2.  RE: Benefits of Tunnel Hub Servers in large environment

    Broadcom Employee
    Posted Jun 28, 2022 01:16 PM
    Usually, believe it or not the biggest benefit is that IM will have better access to everything as communication goes through 1 port usually 48003, and not every port the thick client needs. I only set up Tunnels when there are port restrictions to and from a subnet where hubs need to be placed and IM needs to work. With a tunnel you can reduce all UIM traffic TCP and UDP over 1 TCP port with the  Tunnel Client initiating connection if a direction needs to be specified.

  • 3.  RE: Benefits of Tunnel Hub Servers in large environment

    Posted Aug 07, 2022 10:10 PM
    The main reason we use Tunnel Hubs is availability, and load management. Using a Windows based hub, you are limited to around 20 or so remote hubs\tunnels before you start seeing issues with connectivity - tunnels dont start, or stop and restart, queues fail to connect, etc. This is a limitation of the Windows IP stack, as it switches to a round-robin allocation when too many connections are present, instead of a FIFO queue.

    We deploy two tunnel hubs per 20 remote hubs - the tunnels are up from both, but the queues only go to one - we use the HA probe on the "secondary" tunnel hub to fail the queues over in the event the primary tunnel hub is down (for maintenance, OS patching, etc).