Hi there! I’m currently testing what RB2011UAS-2HnD can do for me. I’m actually dissapointed.
The scenario is: left side RB2011 - right side RB912UAG-5HPnD
As seen on picture this is udb bw test, I expected more, and ufcorse no lost packets.
And this is tcp test, with no packets loss.
So what I did. I reset both boards on factory settings, with no default, so I just set what needed to connect both RB’s to gbit port, and did this test. I have eth2-5 in switch1, so I guess this should work as max as port can do.
Testing between routers limits the cpu power because it needs to generate the trafic. It has been said thousands of times.
So your test is meaningless.
Well I don’t agree with what you have said. CPU should be on 0%, or maybe couple of % for generating traffic. RB912 is on ether port which’s master port is in same switch (switch1) so it should go WIRE speed. Just look at the rb2011’s diagram. I dont believe it’s “passing” hardware chip inside the RB.
Edit 1: So if I use btest.exe on windows should it go gig? It goes even worse. about 10-20mbits
You don’t have to agree…it’s a fact. You are wrong and you tests sustain that.
The switch helps with the trafic between the ports, but not generated by the cpu.
Also btest on windows is not working as it should.
RB912 switch? Doe it have more than one eth port?
You should understand what am I saying, you won’t reinvent the weel just because you feel that way. And router os and routerboard won’t change because of you.
Again. It is written here many times. Place two computers on both sides of the router and push the traffic through. Make the same test without the router in the middle to be sure you have reached the limits of the router or of the computers.
The RB2011 can easily do 1Gbit full-duplex on all it’s gigabit ports when switched. It’s also quite capable of passing traffic when bridged, especially using the new “fast” methods as introduced in the newest RouterOS versions.
Testing bandwith using the bandwith test on the device and thus it’s CPU is not testing the throughput of the device but only what the CPU can generate. This test is workable for lower speeds links, but certainly not gigabit.
Please take a look at a blog post I made explaining how iperf works. Using 2 PC’s/Laptop’s you can measure the performance as you wish. That way the device will only be responsible for passing the traffic and not actually creating it, as it would also do in a real situation.
Sorry, but they are right. Your own tests show that you are maxing out the CPU for the tests. The built in bandwidth test is fine for a slow link, but not gigabit. The RB2011 series is more of a “home/small office” router. It can handle most every small office situation just fine. I have one on a 50/50 fiber and it works great. The key is that you must understand how your are configuring the unit, and understand what you need it to do.
Example; A bridge is not the same as a switch. Bridging two switch’s will not give you wire-speed. Mangle and Filter rules cause evert matched packet to be processed by the CPU. Unless you are using a CCR, this can be very slow. On the other hand, switch ports, tied to a master port will be very fast, if on the same subnet. This is also why many of the routers have Block Diagrams available. They help you to see how the packets will flow and help you understand what to expect from the router.
I thought I would throw in my thoughts on what I have experienced with the RB2011 router. We temporarily moved a network to a RB2011 to do some maintenance on a Cisco router that was having issues. I was pretty apprehensive about whether the RB2011 SoHo router would be up to the task of supporting ISP grade traffic for the time we needed it to.
I was pleased to say the least that the RB2011 performed awesome. At peak this network passed approx 250/50 Mbps and roughly 18K pps at 30 to 40% CPU. I performed a speedtest in the morning when traffic was around 150Mbps and speedtest resulted in 350 Mpbs up/down. I know that speedtest isn’t a stellar bandwidth testing tool but I wasn’t using it to determine SLA compliance. I was just seeing what the router could pass.
At the time of my test we had zero firewall rules and approx 12 NATMap rules.
So although we have CCR routers on order, in a pinch these RB2011 SoHo routers beyond exceeded my expectations.
There is absolutely no guarantee that you will get the same performance. 350Mbit is actually quite the max of the RB2011 before the “fast” methods where introduced. You can use those, but not combined with some features, so the real answer is a bit more complex.
The best is to do your own bandwidth tests and then learn to understand why you are getting certain results and try to figure out what you should get.
In general I would say the RB2011 is fine for gigabit switching, 600Mbit routing or 200Mbit NAT internet, or a mix of those depending on rule and if you are using “fast” methods or not.
Yeah test above is irrelevant. Live scenario was a “good test”, but yeah I thought I could get 1gbit throug it, and I didnt. That was the conclusion. We both bumped in old topic
It goes even worse. about 10-20mbits
