How does the Persistence Load Balancing Algorithm determine WAN selection?

General
1

Hi everyone,

I would like to better understand the detailed logic of the Persistence load balancing algorithm in Peplink routers.

According to the Peplink documentation, Persistence can route traffic through the same connection based on source and/or destination IP addresses, and keep the traffic on the same connection until the session ends.

I would like to clarify the actual decision logic behind this algorithm:

  1. What is the actual decision key used by the Persistence algorithm?
    For example, when selecting Source IP, Destination IP, or Source and Destination IP, is the persistence mapping based only on IP addresses?
    Or does it also include protocol, source port, destination port, TCP/UDP session information, or other flow attributes?

  2. Is Persistence a per-session decision or an IP-based persistent mapping?
    For example, if a client 192.168.1.100 accesses an HTTPS website 203.0.113.10, and the rule is configured with Source IP persistence, will all subsequent sessions from 192.168.1.100 use the same WAN until the persistence entry expires?
    Or is the WAN selected independently for each new TCP/UDP session?

  3. How long does a Persistence entry stay active?
    Is the persistence entry removed immediately when the TCP/UDP session ends, or does it remain for a timeout period?
    Is this timeout configurable?

  4. For HTTPS web browsing traffic, if Persistence is configured based on Source IP, does the router bind the internal client IP to a WAN connection regardless of the destination website?

    Example:
    Client: 192.168.1.100
    Destination A: 203.0.113.10
    Destination B: 198.51.100.20

    Will both destinations use the same WAN because the Source IP is the same, or can they be assigned to different WANs?

  5. What happens to existing Persistence mappings when the selected WAN becomes unhealthy or disconnected?
    Will the router immediately reassign the affected traffic to another healthy WAN?
    After the original WAN recovers, will the traffic return to the original WAN, or continue using the new WAN until the persistence entry expires?

  6. When using Persistence with multiple WANs that have different public IP addresses, does it guarantee that the selected traffic will keep using the same WAN public NAT source IP unless failover occurs?

The reason I am asking is that I want to understand whether Persistence is mainly IP-based, session-based, or based on a 5-tuple flow logic.

Thank you.

2

I think the simplest answer is that “Persistence” policy is just the “Weighted Balance” policy/algorithm BUT with awareness of existing connections before assigning a WAN to a new connection. If there are no existing connections (either “by source” IP or “by destination” IP) then “Persistence” is just the “Weighted Balance” algorithm.

So with that being said, let’s establish a baseline: how comfortable are you with your understanding of the “Weighted Balance” algorithm?

PS – In practice, people only use Persistence for HTTPS due to old/buggy servers that use the source IP address for session tracking or a clumsy attempt at a security sanity check.

1 Like