From help-file :
TEQL queue CONFIG_NET_SCH_TEQL Say Y here if you want to use the True Link Equalizer (TLE) packet scheduling algorithm for some of your network devices or as a leaf discipline for the CBQ scheduling algorithm. This queueing discipline allows the combination of several physical devices into one virtual device. (see the top of net/sched/sch_teql.c for details). This code is also available as a module called sch_teql.o ( = code which can be inserted in and removed from the running kernel whenever you want). If you want to compile it as a module, say M here and read Documentation/modules.txt.
From source-files :
How to setup it.
----------------
After loading this module you will find a new device teqlN
and new qdisc with the same name. To join a slave to the equalizer
you should just set this qdisc on a device f.e.
# tc qdisc add dev eth0 root teql0
# tc qdisc add dev eth1 root teql0
That's all. Full PnP 8)
Applicability.
--------------
1. Slave devices MUST be active devices, i.e., they must raise the tbusy
signal and generate EOI events. If you want to equalize virtual devices
like tunnels, use a normal eql device.
2. This device puts no limitations on physical slave characteristics
f.e. it will equalize 9600baud line and 100Mb ethernet perfectly :-)
Certainly, large difference in link speeds will make the resulting
eqalized link unusable, because of huge packet reordering.
I estimate an upper useful difference as ~10 times.
3. If the slave requires address resolution, only protocols using
neighbour cache (IPv4/IPv6) will work over the equalized link.
Other protocols are still allowed to use the slave device directly,
which will not break load balancing, though native slave
traffic will have the highest priority.