Sorry if this is repetitious. I’ve seen a couple of threads that partially ask the question from other angles, but I want to ask it very clearly:
Under what circumstances are the user-specified upstream and downstream bandwidth figures taken into account when load balancing? As I understand it they are only when using:
- Persistence (if load distribution is set to “Auto”, load is balanced based on available bandwidth)
- Overflow (downstream bandwidth used to determine whether a given connection is saturated before overflowing)
- Least used (as I read it, it’s the one with the most spare bandwidth).
The weighted balance algorithm ignores the specified bandwidth values.
Is this correct?
Also - to clarify “Least Used”: I have two WAN connections, a 5Mb/s ADSL and a 50Mb/s Fibre connection. The Fibre connection is carrying 10Mb/s of traffic from seven clients, whereas the ADSL is idle. Will a new request will still be routed to the Fibre connection because it has 40mB/s capacity?
The illustration shows percentages of capacity. Does it actually use percentages, or absolute capacity? I’d imagine in the real world it would be more useful to use absolute capacity, but then you might need to rename the algorithm to “available capacity” or similar.