1100AHX2 x2 throughput.

Hi,
At one of our sites we have 2 1100AHx2 running as we need the extra ports and rather than go back to L2 with a switch I put another RB in there and set up a /30 between them both.

Performance, however, has taken a hit. If I bandwidth test from 1 RB to the other UDP I get close to 1Gbit/s however if TCP it’s more like 200Mbit/s. I wasn’t expecting TCP to be as high as UDP but to only achieve a fifth of the throughput doesn’t seem right.

The two RBs are connected with a CAT5 patch cable which I’m going to swap out today for a shielded CAT6, but if it was electrical problems surely the UDP wouldn’t achieve such a high throughput.

I’m a bit confused here…any ideas?

Adyb

Never use BTest from one RB to the another. Alway BTest through the RB.

The BTest itself consumes A LOT of CPU to generate and handle the traffic, so if you BTest from one router to another, the results will be very inaccurate. Btest with 2 PCs on both sides of the RBs and you will see very different results.

You can see this very easily when you open “/tool profile” when running BTest. Notice that the BTest process itself is taking most of the CPU, rather then routing/firewalling/queueing.