WAN trunk on a USG1100 firewall - balancing question

Matt10669
Matt10669 Posts: 20  Freshman Member
First Anniversary 10 Comments Zyxel Certified Network Administrator - Security
Hi, I've got a Zyxel USG1100 in my LAN, connected to the web with 2 different internet connection (port P1 and P2 on the firewall). I want to use the two connection at the same time in order to have the max bandwith. I've configured a WAN trunk and I have tested all the 3 algorithm (WRR, LLF and Spillover) but as you can see in the image attached, it seems just the first connection (P1) is used intensively. Is it normal? What is the right algorithm if I want to use both the internet connections at the same time and in the same way?

All Replies

  • USG_User
    USG_User Posts: 369  Master Member
    First Anniversary 10 Comments Friend Collector First Answer
    edited December 2021
    What do you expect? An identical utilization of both WAN lines? Or a duplicated internet speed which summarized both single line speeds? For example, if both WAN lines have a capacity of 100 Mbit/s, do you expect a total bandwith of 200 Mbit/s? It doesn't work in this manner from my point of view. As long as WAN 1 is fine and fast there is no need to devide and distribute the traffic in equal parts to both lines.

    We are using 2 different WAN speeds. That's why the seconds slowlier line is our failover line only, not in use as long as WAN 1 is working fine.
  • Matt10669
    Matt10669 Posts: 20  Freshman Member
    First Anniversary 10 Comments Zyxel Certified Network Administrator - Security
    edited December 2021
    Thank you for the reply. The connections are identical (same ISP, same speed). So you mean as long as the 1st line is not overloaded the 2nd line is not used? At the moment I've configured the LLF algorithm. Do you think this is the right choice?
  • USG_User
    USG_User Posts: 369  Master Member
    First Anniversary 10 Comments Friend Collector First Answer
    edited December 2021
    There are different opportunities in place how to configure a WAN trunk consisting of two "internet" lines. But no of them summarize both bandwiths in general. We are using "spill-over" setting only, where the 1st line (100 Mbit) is permanently active while the 2nd line (50 Mbit) is passive and will be actived only when the 1st line is failing. We don't want to use the second line simultaneously since this slower DSL line has worse ping times in relation to our fiber line.

    By the way, BWM is disabled with us. All settings are made in Network > Interface > Trunk.

    Alternatives to "Spill-over" are "Least Load First" where USG is balancing the outgoing traffic between both lines in accodance to the current utilization of one line.
    Or finally "Weighted Round Robin" where the USG is scheduling service queues on a rotating basis for each line, means sending traffic through each WAN interface in turn.
    For more info please refer to the manual around page 428, "Load Ballancing Algorithms"

    But since we are using Spill-over only, I have less experiences in the other opportunities. Finnaly you have to try the different opportunities to find our that one which suits your needs best.
  • imaohw
    imaohw Posts: 123  Ally Member
    First Anniversary 10 Comments Friend Collector First Answer
    @Matt10669 - the more clients you have accessing the internet the more balanced the traffic will appear.  An individual client will “stick” to a single WAN interface.  Multiple clients will be balanced across WAN interfaces.
  • CHS
    CHS Posts: 177  Master Member
    First Anniversary 10 Comments Friend Collector First Answer
    I Agree. "More clients you have accessing the internet the more balanced the traffic will appear".
    Because port usage is only prove upload/download traffic usage, but can not prove "forwarded session".
    User1 can initial 50 sessions but traffic less than 3M.
    User2 can initial 1 session but download 200M file. 

Security Highlight