Actual line speed calculation

[pestebogdan]

Hi guys,

Perhaps you could help me with an answer to the following predicament:

An SP rented us an MPLS line of 2 Mbps.

Through the line stress tests we came out with 1.85 Mbps.

He states that this is normal due to header information.

150 kbps lost only due to header information ??? (I should mention that the 1.85 Mbps value is taken from the show interface command).

Is he yanking my chain or is there any truth to this ? Maybe I'm missing it.

Thanks,

Bogdan.

[paolo bevilacqua]

Yes, what you have been told to you is correct. However to measure speed, the show interface value is of little significance.

You can spend time measuring, etc, and the end results will be that speed of this 2mbos circuit is exaclty 2mbps.

[pestebogdan]

Thanks for the quick answer,

I actually filled up the bandwidth and kept it filled for about 15 minutes when I did the show interface and it reported the 1.85 Mbps.

I also did an end-to-end test with iPerf and it reported the same values.

Could you please explain (if you have time of course) why his calculations are correct ? How are packet headers causing a 6.5% drop in actual traffic, because I just can't seem to get the math down.

Thank you so much,

Bogdan.

[paolo bevilacqua]

Which exact interface type is that? You are not providing enough information.

Anyway, you can search for "link overhead" to learn on the matter.

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Much depends on how your stress test was generating traffic.

Assuming you were generating just IP packets using full size (std) Ethernet, 92.5% (1.85/2) seems a tad low.  I would expect around 96% (1480[1500 - 20 for IP]/1542[Ethernet overhead plus MPLS label]).

If the physical link is "faster" than 2 Mbps, and is being policed, policer might be "rounding" down "over rate" packets enforcing the 2 Mbps.

[pestebogdan]

Hi,

Thanks you for the answer, but I am still a bit confused as to why did you take into account the MPLS label ?

When the provider rents me a 2 Mbps line i expect 2 Mbps coming out of the PE router (outbound interface to my CPE), not the interface to his P routers.

So it would be his duty to take into account the MPLS label overhead and provision an MPLS link of more than 2 Mbps in his network in oder to deliver a 2 Mbps line to me, right ?

Thanks again,

Bogdan.

[Joseph W. Doherty]

Disclaimer

The  Author of this posting offers the information contained within this  posting without consideration and with the reader's understanding that  there's no implied or expressed suitability or fitness for any purpose.  Information provided is for informational purposes only and should not  be construed as rendering professional advice of any kind. Usage of this  posting's information is solely at reader's own risk.

Liability Disclaimer

In  no event shall Author be liable for any damages whatsoever (including,  without limitation, damages for loss of use, data or profit) arising out  of the use or inability to use the posting's information even if Author  has been advised of the possibility of such damage.

Posting

I took into account the MPLS label because often when a provider provides # Mbps on a physical link, they emulate a physical link of # bps.  In other words, if they provided you a E1 of 2 Mbps, the circuit provides 2 Mbps of raw capacity, not 2 Mbps of payload or IP capacity.

Let's also assume even if they give you a Ethernet or FastEthernet handoff, but their backside link is using an E1.  Bandwidth is going to be lost for the MPLS label.

On the other hand, if your frontside link is an E1, and not natively MPLS, and their backside link is something "faster", they could not account for the MPLS label bandwidth on the front side, but then they have a problem if you upgrade the frontside to match their backside bandwidth.

What the provider's actual policy is, is up to your provider.  The above, hopefully, explains why they might consider it part of your bandwidth even in situations where technically it doesn't need to be.

BTW, it would help if you would clarify whether this 2 Mbps is a logical cap or a physical cap.  I've been assuming it's a logical cap.

If it's a logical cap, I've also assumed the provider polices or rate-limits, but they don't shape.  Even if we exclude MPLS label overhead, your measured rates still a little slower than expected, which is why I'm thinking packets that are "fractional", i.e part of a transmitted packet exceeds the bandwidth cap and the whole packet is dropped (including the portion that wouldn't be over rate).  I.e. you might see an utilization difference for difference size packets (after accounting for their different overhead requirements).