Hello,

I made a MPLS connection (on RB 750) with this TUT (http://wiki.mikrotik.com/wiki/Transpare ... using_MPLS) but did not use wlan1. I used wired port 5 instead.
Now when I make a RFC 2544 Test with EXFO FTB-1 the troughput is right only with the size of 1518.
The throughput with 64, 128, 256, 512, 1024 and 1280 is not much enough. As example 10 Mbit/s with size of 64 (should be nearly 100 Mbit/s)

Can anyone help me please?

Regards
Kangoo

This is normal, you will never get the same BW with small packets. Look at CPU usage and you will see why.

I get this result:

size throughput in mbit

64 16,9 Mbit/s
128 30,6 Mbit/s
256 64,00 Mbit/s
512 78,90 Mbit/s
1024 98,70 Mbit/s
1280 99,38 Mbit/s
1518 100,00 Mbit/s

Cpu usage is low. But what is the total throughput now?
As example a customer needs a 50 Mbit/s L2 connection. Will this than be OK?

Regards

Kangoo

As customer, if I specifically tested small packet sizes, I would not accept that performance, as I ordered 50 Megabit/s, and I should be able to reach that performance regardless which packet size my preferred application uses.

That said, as ISP you could probably argue with IMIX (http://en.wikipedia.org/wiki/Internet_Mix) or 512-byte performance, as there are not a lot of use-cases (VoIP/SS7 would be some of those) where packet sizes are lower than that. It kind of depends on what you promise/guarantee in your terms of service.

Also take care to correctly set MTU/L2MTU/MPLS MTU to their respective correct sizes as to prevent fragmentation (which really slows your devices down).
However, if you do value your customers, please give them something better/faster than a 30$ CPE for a virtual leased line...

OK, The Connection with the RB750 only was for testing.
Which RB Model would be nice to give a custumer a correct 50 Mbit/s MPLS/VPLS connection confirm with RFC 2544?

Regards
Kangoo

OK, The Connection with the RB750 only was for testing.
Which RB Model would be nice to give a custumer a correct 50 Mbit/s MPLS/VPLS connection confirm with RFC 2544?

Regards
Kangoo

Compare the routing packet performance on routerboard.com. This should tell you what board is fast enough.
I would go with a RB1100AHx2.

The thing is - according to official performance figures, RB750 should be fast enough to handle 50 Mbit/s in 64 byte packets, as MPLS performance is supposed to be somewhere between bridging and routing, but I think Fastpath does not work with VPLS. The safest choice of Routerboard for this application would be a RB1100AHx2 then, but it will scale to 100 Mbps and higher, if required in the future.

Just be very cautious to correctly use the port groups on the RB1100AHx2, as for example moving traffic between ether1 and ether2 will be way less effective (as in: slower, use more cpu) than moving traffic between ether1 and ether6. You can find a flow-chart and recommended port usage pattern somewhere in the forum. In general, you should avoid moving traffic between members of the same switch-group (ether1-5 and 6-10), unless that traffic is actually switched (by using switch-chip functionality).

The thing is - according to official performance figures, RB750 should be fast enough to handle 50 Mbit/s in 64 byte packets, as MPLS performance is supposed to be somewhere between bridging and routing, but I think Fastpath does not work with VPLS.

This applies only to LSR. Using RB750 as LER will give lower numbers and using it additional as VPLS endpoint will cost another cpu cycle.