Compression question: what do users normally use layer 2 or layer 3? Or what is easier to configure? Does layer3 give you more bw or it's all same just that layer 3 must only be ipv4. If I'm transmitting at 256 Mbps with no compression what would the speed be with layer 2 and layer 3? Thanks,
It really depends on your network design which mode you can use. L3 is the preferred mode as you get the intelligence, but it requires that you only have IPv4 traffic on your network. No L2 information is carried, i.e. no vlan tags, no MPLS traffic, etc. If you have any need for non IPv4 traffic, you'd need to use L2 mode.
The intelligence provides expanded throughput in L3 mode, so you'll always get a higher effectiveness with L3 than you would with L2.
As the actual compression effectiveness is going to depend on your actual traffic mix, it is hard to say for sure what effective throughput you will see, but there are some rough numbers in the L2/L3 guide. Numbers under 1 (100%) indicate that the compression is not producing a benefit. Numbers over 1 are yielding you a benefit. You can take the number and multiply it times your raw RF throughput to come up with the approximate effective throughput, i.e. 256 x 1.29 = 330 Mbps.
If you switch to L3, please note the additional step needed to change the dtag setting covered in the setup guide.