Trango Atlas vs MT on RB532 and nstreme

Can someone tell me why there is so interesting difference between these two solutions:
Both were tested on the same link with real traffic up to 16Mbit:

Trango Atlas:

  • excellent latency - up to 18ms
  • max p/s up to 2200 on graph
  • high packet lose ( on ethernet ports when high traffic, ping form radio to radio no packetlose )

MT on RB532 with nstreme and CM9 in 5GHz Turbo

  • bad latency - up to 50-100ms
  • max p/s over 3500 on graph
  • no packet lose

The MT is great but problem is the latency. Is it due to CPU utilization near 100% ?

  1. I’m surprised the Atlas only did 16mbps. Was your RF path clear?
  2. The MT retransmits at layer2, so you see latency instead of packet loss.

16Mbit is the real level of traffic, it can transmit more. The problem of Atlas is in packet lose on ethernet interface !!! My situation:
—ethercable——wireless— —ethercable—

When I ping from router1 to atlas1 or router2 to atlas2 during high traffic utilization, there is packet lose. When I change Atlas to MT, there is no packet lose. Therefore it do not depend on wireless link I think.

And can you advice how to check quality of Layer2?

how were you testing throughput?

throughput is not important, testing is done on real traffic. But when I want to test throughput, I use bandwidth software from behind routers.