I have a question if i want a make a pppoe server and i have a 600 connection is the routerboard 1100 AHx2 keeping good or i will have a problem with this big traffic and with this a lot of people ?


thank you

The question is how much bandwidth do you want to give your users?

30 mbps or 50 mbps per user

per USER??? 8-j
600*30Mbps.. it's around 18Gbps

I think, this router not do the the full 10Gbps routing at all 13x1Gbit ports with pppoe+bandwidths(queue+mangles).
As example: 250Mbps channel (full duplex in+out=500Mbit) (average load 60-70% - total bandwidth around 300-400Mbit) eat fully (average cpuload 75%) the dual core router (E8600) at 3.3Ghz with 2 intel 1Gbit-netcards - around total 2500 users (at one moment to 1k users), without pp* - only nat+queuetree(dynamic ~1k-1.5k rules - 2 rules per user)+mangles (~1k-1.5k rules). 25-30% of cpu-usage eating the mangles? 5-10% eating by queue and other - input-output. And 1100 AHx2 have less power for route and compute, but if you can use pcq-type queue (we can't use it), when one queue-rule for many users, and no many mangle-rules - then.. may be this RB1100AHx2 can do full pps

You want to got for x86 with this amount of users and needed bandwidth per user.

I am already maxing out he CPU of a RB1100 (not AH nor AHx2) with about 100 PPPoE users and up to 30Mbps each.
Conntrack is off and one static MSS Mangle rule is used to optimize the CPU load.
The Router just does PPPoE. Everything else is offloaded to other routers.

now lets say a RB1100AHx2 would be three times faster than the RB1100.
Then do the math yourself...

This is strange a little bit users for this hardware and software.Alright how many traffic is the maximum for this router (1100ahx2)?

for larger users its more recomender to use external radius server, because the limitation in the routeros database.

This is strange a little bit users for this hardware and software.Alright how many traffic is the maximum for this router (1100ahx2)?

An example - one RB1100U can thru around 10-15Mbit per each 10% cpu usage with queue, nat, static routes. RB1100AHx2 more powerfull - *2 than simple (old) RB1100U - so, 25-30Mbit per each 10% of cpu. It's around and example, because I see this ratio (between cpu usage and traffic thru) with no PCQ queue-type. May be, when using one PCQ-type queue for many users, than CPU can thru more traffic.

I think, that in real traffic this RB1100AHx2 can do thru around 150Mbit full-dulex channel without any problem, but for 200Mbit it may be not

Of cause, if you have 100Mbit channel, when you can do many bands for 5, 10 or 30Mbit because not all users work at one time.