First, is there a manual somewhere that explains this, please let me know.
If I have 2 active connections in the same priority, Priority 1, how does peplink decide which to use? Round robin, based on ping, etc? This scenario is not considering SpeedFusion.
What type of priority are you asking about? QOS or outbound policy?
I am pretty sure all race conditions are handled with FIFO (QOS).
Two active connections for outbound policy would go to the same WAN if it is still healthy.
Thanks. I’m looking at he dashboard where you can drag connections between priority 1, 2, 3, 4, and disabled.
From you notes it sounds like behavior depends on how the inbound and outbound policies are setup. If fifo is the method used, does this mean the first connection drug into the priority is used first? Sorry, I’m not understanding.
What device are you talking about?
Typically, priority is used for fault tolerance as opposed to bonding or throughput. As long as the connection with the highest priority (lowest number is highest), traffic will continue to use that link. If the link fails a health check, that is when the link with priority 2 will be used for new connections.
This information is assuming that you are talking about outbound policies, but those aren’t visible in any of my dashboards, so I could be way off. If it is a SOHO or similar cellular based router/appliance - I don’t have any to play with. I could see how that might be useful for assigning WAN connection startup/availability based on other factors. I.e. prefer WiFi as WAN as 1st, then use simA if it isn’t available. I am pretty sure the mobile-y devices have such feature sets.
I am speaking from a Balance series router user perspective.
By default on a MAX device all WAN connections in the highest priority use weighted load balancing with equal weights.
You can change how that works in outbound policy.
I checked the outbound policy on my unit. It is set to “persistence” by default. This is likely why I’m not getting any use out of the 2nd connection in the same priority. I’ll change this to weighted and monitor the results.
Very unlikely. I suspect you missed the far from obvious link on the outbound policies page:
Which then shows this page that in turn highlights I’m wrong. Default policy (when set to auto) is lowest latency on a MAX:
You can either edit that default policy or add a new one (I always add a new one). If using weighted load balancing make sure you set the weighting to ‘1’ for each connection you want to load balance rather than ‘10’.
Oh! Thank you so much for pointing this out. Indeed I think I did overlook the default.
Hey Martin. Will you please share what default policy you use and why weighting to ‘1’ is important? Thanks much.
I have two fiber connections, so I use lowest latency.
For even load balancing. By default its set to 10. This means that the first 10 outbound sessions will be sent via WAN1 and the next 10 via WAN2. I’m sure that works will in larger office environments with 20+ users but for smaller user / session counts you more likely want a round robin approach of 1 session on WAN1 and the next on WAN2 etc. Hence why I set it to 1 instead of 10.
