I’ve tested with NAT/Masquarade through L2TP tunnel on WAN. 600MHz clocked RB2011 is 30% faster than 680MHz clocked RB450G. 25 rules in firewall, P2P traffic. If RB2011 is overclocked to 750MHz the performance difference is near 40%.
but sometimes 2x more RAM helps RB450g a bit ;=)
btw, what best stable clocks for 2011UAS “in general” ?
anyone experimenting with stock setup ?[no extra radiator/fan, no powe system update].
I’ve done some real-world measurement with RB2011L, RB493G (should have the same performance as RB450G) and RB750G, here are results:
RB2011L-IN 6.0rc11
routing: 914 / 901 Mbps
firewall + conntrack: 464 / 478 Mbps
NAT: 360 / 355 Mbps RB493G 6.0rc11
routing: 624 / 574 Mbps
firewall + conntrack: 418 / 428 Mbps
NAT: 352 / 362 Mbps RB750G 6.0rc11
routing: 450 / 408 Mbps
firewall + conntrack: 243 / 226 (just one time 276) Mbps
NAT: 217 / 186 Mbps
But I have serious troubles with RB2011L - despite the fact I am using it at default 600 MHz, it is very unstable when the traffic is near 900 Mbps - it stops to respond and starts to act as switch. I have to turn it off and on again I will do some more testing (maybe with lower frequency?) and the open new thread about that.
one question about this routers performance
i need a 300mbit transfer for 5 km
what do you say about: 2x 5Ghz dual links (with SextantG) and a cheap solution for bonding this 2 links transfers
in the @zervan’s previous post sad: RB2011L-IN 6.0rc11
routing: 914 / 901 Mbps
firewall + conntrack: 464 / 478 Mbps
NAT: 360 / 355 Mbps
what do you say, can handle the RB2011L-IN and bounding the 2 links in 1?
in v6rc14 we have included new RouterBOOT version for RB2011, it fixes such problems. Please try upgrading to rc14 and new RouterBOOT and see if you have similar issues ever again
I’ve upgraded now (I was avoiding it before because I use VLANs and there is a bug in v6rc14 relating to it, see http://forum.mikrotik.com/t/v6rc14-released/65819/45) and as I was expecting, I’ve lost connection - ether1 didn’t receive any packets since then
I am using ether1 as the only port for LAN1 (my side), switched ether2-5 ports are bridged to ether6-10 switch as LAN2. I want to use VLAN on all the ports, but there was a bug in v6rc13 in switch2 (see http://forum.mikrotik.com/t/rb2011l-dhcp-not-working-on-eth6/58426/8) - you have fixed that in v6rc14 but also broke something in switch1. If fallback VLAN mode is set on switch1 cpu, then ether1 is not receiving anything and it doesn’t depend on VLAN mode for ether1. Well, I’ve set VLAN mode to disabled on switch1 cpu and it started to work. But ether2-5 ports didn’t “see” my VLAN packets then - moreover, they were not receiving any packets with VLAN tag, I have to set them to disabled VLAN mode as well…
Please, do more testing. It is not normal when a fix of one bug will create another different bug.
I think RB2011 should be enough powerful to bond these two links together, but I have never tried that. In fact, I haven’t ever need interface bonding.
It seems like it doesn’t matter - there is RB250GS switch connected to ether1, but when I was looking what’s wrong, I took RB2011L and connect it to other devices (RB1200 and maybe computer directly too, I don’t remember).
I was using iperf with 5 parallel transfers, TCP Window 64 kB, buffer 1 MB. It corresponds best to my needs - practically the same performance as when I make multiple transfers from server / NAS to client PCs.
rb2011 in my tests is stable with v6rc8(after loading firmware, one freeze, power cycle and works 2months without problem. Interesting that every upgrade to rc8 it freezed once and after power off it works.) I tryed rc11,12,13 many freezes, reboots.
Thanks for test. 1. on the internet.