Alien Wave service deployment. Challenges going from experiments to live production. Roeland Nuijts, SURFnet Lars Lange Bjørn, NORDUnet Martin Nordal Petersen, DTU Fotonik Anna Vasileva Manolova, DTU Fotonik Tijmen Van den Brink, Telindus Magnus Bergroth, NORDUnet Rasmus Lund, NORDUnet GN3 JRA1 & 2, Workshop 20 – 22 Nov Kastrup, Denmark connect • communicate • collaborate 1
Outline Introduction to the project Drivers/Motivations Alien Wave concept Past simulations, experiments and conclusions OAM&P Operational aspects of productizing the Alien Wave Monitoring concepts. Performance Next steps connect • communicate • collaborate
Introduction to the project Idea Idea - Deploy alien wavelength between 2 NREN’s using - Deploy alien wavelength between 2 NREN’s using different vendors different vendors Hero experiment Hero experiment - Establish the aw and show proof of concept - Establish the aw and show proof of concept Testing Testing - Experimental results for knowledge, optimisation and - Experimental results for knowledge, optimisation and possible guide lines for future AW deployment possible guide lines for future AW deployment Verification Verification - Simulations for verification of the experimental results - Simulations for verification of the experimental results Production Production - Establish procedures, agreements and a verified end - Establish procedures, agreements and a verified end product product connect • communicate • collaborate
Drivers Direct connection of customer equipment to a third party DWDM equipment and elimination of expensive transponders at the transition between two DWDM systems, Elimination of regeneration reduces power dissipation Faster time to service cost savings Support of different modulation formats offers extended network lifetime. connect • communicate • collaborate
Alien Wave concept RX TX TX RX Conventional DWDM RX TX RX TX RX RX TX TX Multi domain DWDM systems TX RX RX TX Mulitdomain DWDM systems with alien wavelength connect • communicate • collaborate
Past simulations, experiments and conclusions • 10G neighbours • Guardband variation 50 – 150 GHz • Full control of spectrum in NORDUnet • Off-line VPI simulations connect • communicate • collaborate
Past simulations, experiments and conclusions • Lower number on y axis represents higher BER. • XPM effect seen with higher intensity of the neighbour channel power VPI Results: 10G variation w. constant 40G connect • communicate • collaborate
Past simulations, experiments and conclusions • Simulations done with 10G off to find SPM (if any) • Initial better BER with higher intensity, but then SPM kicks in. (7dBm) connect • communicate • collaborate
Past simulations, experiments and conclusions Optimum area? 40G variation w. constant 10G reveal both XPM and SPM connect • communicate • collaborate
Past simulations, experiments and conclusions Experiments backup the simulations and, “near optimum” area can be investigated with guard band variation testing (Graph@ 50Ghz spacing) connect • communicate • collaborate
Past simulations, experiments and conclusions 50HGz to 150GHz @ 10G = - 20,7 dBm The pre-FEC BER is below 10 -3 unless spacing is above 100GHz The Alcatel-Lucent design guide specifies 10G = - 17,2 in order to maintain guarantied performance according to traffic matrix! connect • communicate • collaborate
Past simulations, experiments and conclusions Cost savings 200G Amsterdam - Copenhagen • Regenerated CAPEX cost • Cost of equipment is the arround 50% higher predominant factor by orders of compared to the AW CAPEX magnitude (CAPEX) • Operational cost can be left out of conclusion (OPEX) connect • communicate • collaborate
Past simulations, experiments and conclusions • Performed simulations conformed qualitatively with experimental results (VPI tool) • Major XPM effects evident when guard band is decreased and neighboring channel power increased • Equipment represents more than 98% of cost for an alien wavelength. In the case with regeneration, costs nearly double. • OAM&P is adequate for this mix of products, few steps are needed for “normal operation” connect • communicate • collaborate
OAM&P Consequence NREN ISSUES NREN TARGETS • No real compatibility for • Full optical transparency newer LH modulation • Bitrate schemes. • Modulation format • No standardization targets • Spacing LH/ULH applications • Full optical control • No standardization targets • Power OTS/OMS interworking • Spectrum • Admission connect • communicate • collaborate
Operational aspects of productizing the Alien Wave Involve respective NOC’s at startup fase Apply necessary staff training for handling AW services Agree on responsibilities and procedures Define operational handover criterias connect • communicate • collaborate
Responsibility agreements Tasks Responsibility 1 st step NORDUnet SURFnet Project partner Create a responsibility Monitoring point (S ‐ >N) Monitoring point (N ‐ >S) matrix with a clear Informing about problem in (S ‐ >N) overview on “who’s Informing about problem in (N ‐ >S) Keeping SURFnet spares for SURFnet in CPH. doing what” and Replacing and reseating boards contact info when Documentation for onsite support Clear assignments for onsite support something goes wrong Spare part inventory tracking Ticket system information exchange Partner Contact info Alarm Situation Telephone Email address +46 8 20 78 60 247@nordu.net NORDUnet +31‐30‐247 7586 noc@surfnet.nl SURFnet Partner Contact info (Non urgent) Telephone Email address +46 8 20 78 60 noc@nordu.net NORDUnet +31‐30‐247 7586 noc@surfnet.nl SURFnet connect • communicate • collaborate
Operational aspects Second step. Define the internal processes for handling alien waves in the live production network Define monitoring points and interpretations across the network Find technical solution to monitor the identified points Test the monitoring setup connect • communicate • collaborate
RX TX CPH connect • communicate • collaborate 3 4 OAM Identifying monitoring points 2 5 HB 1 6 ASD TX RX
RX TX CPH connect • communicate • collaborate 3 4 OAM Identifying monitoring points 2 5 HB 1 6 ASD TX RX
OAM Details NORDUnet monitoring point HMB ‐ ORE ORE HMB OADC1750 Port X TDMX Alarms from individual ports OADC1750 (e.g. OADC 1750) Port Y TDMX shall be masked and relayed toward the Management (Nrx) NORDUnet receive monitoring point system connect • communicate • collaborate
Monitoring solution NORDUnet Nagios script initiates a poll by asking the Monitoring server for alarm status The Monitoring server then polls the Alcatel NMS via SNMP Monitoring server Alcatel NMS connect • communicate • collaborate
Monitoring solution NORDUnet The monitoring server search for Alcatel NMS reply's with current alarms on relevant physical ports. alarm list If alarms are present, it correlates Service alarms are forwarded from them with RM alarm data to check if an service manager and physical associated port is ”in service” or not. alarms are forwarded by the Result is sent back to Nagios element manager Monitoring server (E2E-MP) Alcatel NMS connect • communicate • collaborate
Monitoring solution NORDUnet/SURFnet startpoint Both NMS systems report status to 2 separate the Measurement points on their individual alarm status E2E measurement point Ciena NMS E2E measurement point Alcatel NMS connect • communicate • collaborate
Monitoring solution NORDUnet/SURFnet startpoint Alarm status of Via Plugin, own domain and Nagios and the partners end the Zenoss point can be tool can poll monitored alarms from E2E measurement point both MP’s Ciena NMS and thereby have same alarm view E2E measurement point Alcatel NMS connect • communicate • collaborate
Monitoring solution future PERFSonar?? In future, monitoring points can be integrated to the E2E measurement point perfSONAR system Ciena NMS to get identical overview E2E measurement point Alcatel NMS connect • communicate • collaborate
PERFORMANCE Currently, degradation alarms from WSS components are possible A OADC 2x4:1 to retrieve and forward to B the Management system WSS w. ”OSA” AMP C Thresholds can be altered so that alarms arise before actual signal degradation appears connect • communicate • collaborate
PERFORMANCE In theory, SNMP trap polling method can be expanded also to forward statistics on the link performance. (BER) Data will be sent from both parties, to the centralized monitoring system (e.g. perfSONAR) providing inter domain link health status on the AW NB! In the current AW setup BER BER from Alcatel not possible E2E measurement point Ciena NMS BER E2E measurement point Alcatel NMS connect • communicate • collaborate
Recommend
More recommend