Alcatel-Lucent 1830 PSS and 1626 LM® - EANTC: …

1 EANTC Test Report: Alcatel-Lucent Real-World 8 Tbit/s Throughput Page 1 of 2 alcatel -Lucent1830 PSS and 1626 LM Real-world 8 Tbit/s Throughput Test ReportIntroductionIt is not every day that we are asked to support amajor telecommunications operator in verifying thattheir brand-new transport system delivers thecapacity they require. In June 2013 we receivedexactly such a request by Alcatel-Lucent Teletas. Asthe supplier to T rk Telekom, the incumbent operatorin Turkey, they requested that we measure thethroughput capacity of their DWDM systems builtfor the Turkish operator between Istanbul andAnkara - a 540 km cable , being a major link between Europe andAsia, is an ideal location for an east-west exchangepoint.

2 In addition to this trans-continental passagecapability, Turkey s two major cities boost a totalpopulation of 18 Million. With this populationdensity, one can assume that the broadband pene-tration, reported to be million at the end ofQ1 2013 (reference: T rk Telekom Investor Rela-tions), is also centered there. Capacity is DevicesThe transport system constructed by alcatel -Lucentfor T rk Telekom consisted of two products: the1830 Photonic Service Switch (PSS) and the 1626 Light Manager (LM).

3 The client, connected to the1830 PSS (the device that also connected to ourtesters) included two types of interfaces: 10x10 GbEmux coherent transponder (model: 112 SCX10) and100 GbE add drop coherent transponder (model:112 SCA1).For the actual tests all customer traffic was re-routedto other links and the transponders were connectedto an array of testers. One Alcatel-Lucent 1626 Light Manager terminal was installed in each loca-tion, in addition to six 1626 LM in-line ErbiumDoped Fiber Amplifier (EDFAs), as well as threeOptical Add-Drop Multiplexers.

4 32 alcatel -Lucent1830 Photonic Service Switchs hosted a total of160 transponders (divided equally between the twolocations). Half of the transponders were used tocarry the 100 Gbit/s Ethernet signal mapped intoOptical Transport Unit 4 (OTU-4) payload and theother half were dedicated for the 10 Gbit/s engineers configured 80 wavelengthsbetween Ankara and Istanbul. 39 of these wave-lengths carried 10 x 10 Gbit/s and another 41wavelengths carried 100 Gbit/s. In essence thecarrier has fine granularity on the wavelengths theywish to use and is able to activate specific channelsas needed.

5 For our tests, all 80 channels MethodologyOur interface to the Alcatel-Lucent 1830 PhotonicService Switches was an array of JDSU testerswhich the carrier supplied. In preparation for thetest we agreed to use a methodology, used byEANTC in previous tests, to minimize the require-ment on the number of tester ports. We connectedthe tester ports in such a way that each was respon-sible for generating traffic to a number of transpon-ders. This method is commonly called Snake orBack-to-Back.

6 The principle requires careful plan-ning: the first transmitter port connects to the firstreceiver port on the device under test. The rest ofthe connections are looped through the device, untilFigure 1: Alcatel-Lucent 1830 PSS Test SetupEANTC Test Report: Alcatel-Lucent Real-World 8 Tbit/s Throughput Page 2 of 2the last transmits from the device under test connectsto the receive additional challenge was the fact that wewanted to verify that the link really terminated inAnkara and was not a lab experiment.

7 This meantthat we had to loop the test traffic in Ankara andverify that it is really heading from Istanbul toAnkara in the first place. Since the distancebetween the two locations is well-known we calcu-lated the latency and compared our measurementsto our expectations. Our calculations proved to bewithin 20 microseconds of the results, and since wedid not account for the local fiber connections inboth ends we were convinced that the link termi-nated in designing the tester-to-channel mapping wespecifically made sure that two neighboring chan-nels received distinct bit patterns.

8 The idea behindthis was to ensure that there is no crosstalk betweenthe channels. We also evaluated that the configura-tion was correct by disconnecting each channelseparately and verifying that the signal only disap-pears from every other interface. This can also beseen in Figure 1 - the red LEDs indicate the channelfed by one tester while the green LEDs still carytraffic for the adjacent the last step in our verification tests we senttraffic for all channels. We then disconnected thesingle fiber connection that was leaving the building(see Figure 2 of the optical fiber distribution frame).

9 This was a common ITU-T Single Mode fiberthat disappeared from the building in a fiber ductand continued its way through nine amplificationpoints before reaching Ankara. Once we discon-nected the fiber, all traffic disappeared asexpected. It proved that indeed all our 80 channelswere using this one link. With this indication wewere ready for the formal ResultsAfter verifying that the configuration was correctand that all cables were indeed connected in theright order we started an overnight Bit Error Rate(BER) test.

10 8 hours later we returned to find out thatone of the testers recorded frame loss, however, thedevices under test did not indicate that any errorswhere present. We connected the tester to anotherfiber that showed no errors and ran traffic againonly to record that the errors moved with the replaced the faulty tester and repeated the testthree times, each run at 30 minutes withoutrecording any frame loss. Every tester was sendingtraffic at line rate for its respective interface whichmeant that we were successfully generating 8 Tbit/sof traffic on the link to Ankara and verified that the Alcatel-Lucent -provided DWDM transport between T rk Telekom sites in Istanbul andAnkara indeed carried 8 Tbit/s using a single fiberpair.