Ethernet traffic over SDH equipments

Discussion in 'Computing and Networks' started by Luka Boroti, Feb 15, 2013.

  1. Luka Boroti

    Thread Starter New Member

    Feb 15, 2013
    2
    0
    Hi friends,

    I was testing a leased line that our company has and observed that the bandwidth used if transmitted TCP traffic was oscillating to much. If over the line was transmitted UDP the usage was near to the configured one. For the UDP this is normal but for the TCP traffic it was going from 30% up to 70% and oshilating to much. The test was done by Iperf. We have tested another Ethernet lease line that we have over SDH equipments and observed that the bandwidth was very stable even with the TCP traffic. On the Iperf is used for the test only one session so maybe the oscillations are normal.

    What is strange for me is how the Ethernet over SDH doesn't have this kind of oscillation?
    What i wanted to ask is if anyone has experience with the EoSDH and explain how dropping is controlled in the EoSDH?
    Do the SDH equipment have any congestion notification for the end point equipments that are routers and tells them to slowdown when the congestion starts?
    Can anyone tell me in which ITU recommendation this is defined if any?

    I have checked the ITU-T G.7041 about the GFP but is not cleared for me the mechanism.

    Thank you in advance for your help.

    Best regards,

    Luka
     
  2. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    I'm guessing you are in Europe?

    Are you comparing a copper line (E1/T1) to fiber (SDH/SONET)?

    What distance are you moving the signal over, what bandwidth, and what is the endpoint equipment.

    I'm unsure what you mean by "too much oscillation", what type of oscillation, and at what frequency?
     
  3. Luka Boroti

    Thread Starter New Member

    Feb 15, 2013
    2
    0
    Hi thatoneguy,

    Yes our company is located in Europe.
    The test was performed with a E1 line over SDH and 2Mbps line between routers.
    We have tried different distances and the result was the same.The end point equipments were laptops that were transmitting and receiving traffic.
    With "too much oscillations" i mean that the speed that was possible to transmit varies from 30% to 70% of the 2M line. So it was far from stable on the router connection.
    On the other side on SDH it was very stable moving 3-5% max.
    I am interested to understand how SDH informs the end points equipments to slow down if it performs this action and avoid the collision.
    I am aware how the EoSDH works (VCAT, LCAS, GFP) but we were surprised about the result of the tests.

    Thank you in advance and best regards,

    Luka.
     
  4. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    6,357
    718
    The number of reasons fiber out-performs copper are too great to list here.

    Think about what changed:

    Endpoint equipment at both Border and the router itself namely.
    Bandwidth of fiber.

    What was type of E1, and distance? Was there other traffic on the E1? such as phone lines split out, or Frame Relay?

    The 2Mbit bandwidth of an E1 (1.544 MBit T1 here) is rarely fully utilized, frames need to be resent, routing overhead, etc, resulting in about 80% "usable" bandwidth.

    The physical layer in the TCP/IP Stack abstraction is entirely different between E1 signalling and SONET. It's very difficult to point at just one reason the results are better, but there are several. Check the "related articles" out on the wikipedia page for SONET for an idea.
     
    Luka Boroti likes this.
Loading...