Certification for 10G copper cabling Presented by Eric Lie, Sales Manager 1 Introduction This presentation is intended to provide an overview of the 10GBASE-T requirements for cabling The applicable standards to specify the cabling requirements The impact on field testing Fluke Networks introduces an accessory kit for the DTX-1800 to include Alien Crosstalk testing Tests and certify cabling for compliance with 10GBASE-T or new Augmented cabling specifications A review of measurement techniques Description of certification strategy Recommendations What can or should you do today? 2
10GBASE-T What, Where, Why, Who, When? What is 10GBASE-T? A New 10 GbE PHY (OSI-Layer 1) using the existing MAC Where are the 10GBASE-T applications? Initially in the Data Center Early adopters will plan new cabling to support 10GBASE-T in the Horizontal Future proof strategy Why is 10 GbE over copper important? Cost $$$: Port cost is cheap relative to 10 GbE Fiber Who will implement 10GBASE-T products? Both Server and System Vendors (for data & storage) When will it be available? IEEE Standards Board approved 10GBASE-T on June 8, 2006 Network devices: some time after the completion of the standard 3 10 Gb/s Ethernet Applications APPLICATION 10GBASE Fiber (802.3ae) 10GBASE-T 10GBASE-CX4 (802.3ak) Data Center Server Clustering Yes Yes Yes (< 15m) Horizontal In Building No Yes No Vertical Riser Backbone Yes No No Campus Yes No No 4
Customers for 10 GbE Cabling Test Solution Traditional customers for cabling certification testers: Datacom installers/contractors In-house (PNO) organization responsible for cabling infrastructure Benefits: Certify compliance with 10GBASE-T, Cat 6A or Class E A DTX-1800 is a future-proof investment Data Center Infrastructure Staff This group may typically outsource any cabling work We anticipate that the DC IT staff will take a more active interest in cabling issues when planning 10 GbE deployment Benefits: Evaluate whether existing Data Center cabling is ready to support 10GBASE-T Certify new installations for 10GBASE-T readiness 5 Performance Specification for Twisted-pair Copper Cabling Systems Basis for cabling standards: A Signal-to-Noise Ratio Analysis 6
Transmission Performance Analysis Measure Signal strength (or signal loss) over the specified frequency range Insertion Loss measurement (used to be called Attenuation) Must meet specified performance level (less than maximum loss allowed) Identify and measure noise parameters of interest over the specified frequency range Standards define the number of test frequencies (step size) For example: Cat 6 measurement contains 68,712 points/link For cabling certification: Each noise (test) parameter must meet the specified performance level over the total frequency range 7 Cabling Requirements for 10GBASE-T The signal-to-noise budget analysis is complex and consists of two types of disturbances: In-channel disturbances Initially set to ISO/IEC 11801 Class E limits extended to 500 MHz (Class E/Cat 6 defined from 1 to 250 MHz) Certify with the DTX-1800 as is Between-Channel disturbances for Crosstalk referred to as Alien Crosstalk Requires the DTX-10GKIT accessory package Certify with the DTX-1800, DTX-10GKIT and a PC (laptop) 8
Overview of Industry Standards The IEEE standards board approved the 10GBASE-T standard on June 8, 2006 Project 802.3an: 10 Gb/s Ethernet over 4-pair balanced twisted pair cabling Activities in the cabling industry 1. Support for 10GBASE-T over Cat 6 / Class E cabling TIA: TSB-155 (Telecommunications Systems Bulletin) ISO/IEC: TR24750 (Technical Report) 2. Support for 10GBASE-T including alien crosstalk requirements over the full 100 m (horizontal) channel TIA: Addendum #10 to TIA/EIA-568-B.2 TIA/EIA-568-B.2-10 Augmented Cat 6 (Cat 6A) ISO/IEC: 2nd Amendment of ISO/IEC 11801 Augmented Class E (Class EA) 9 10 Gb/s Ethernet Introduces Between-Channel Disturbance Discussion of Alien Crosstalk 10
Alien Crosstalk Crosstalk between wire-pairs in adjacent cabling links General rules: Impact increases with the frequency of the transmitted signals The effect increases with proximity This crosstalk is worst between wire-pairs with the same twist rate The effect is greater for pairs with a lower twist rate Impact increases with the distance over which the cables run in parallel 11 ANEXT and AFEXT Along the Length of the Links Channel configuration as defined Switch C1 C2 Link 1 CP TO PC ANEXT AFEXT Switch Link 2 PC Telecom Room ANEXT is generated in the first 20 m (65 ft) Work Area AFEXT is cumulative through the link and travels its length 12
ANEXT and AFEXT For Varying Link Lengths Permanent link (90m) C2 Victim Link CP TO AFEXT ANEXT Telecom Room Work Area AFEXT influence can be over-powering and cripple transmission on the longer link (victim link) 10GBASE-T must back-off signal strength on the shorter link and a normalization as specified in all standards (IEEE, TIA and ISO/IEC) must be applied. 13 TIA/EIA-TSB155 Draft 5.0 (Installed Base) Additional Cabling Guidelines for 4-pair 100 Ω Category 6 Cabling for 10GBASE-T applications Extended frequency transmission performance of Cat 6 cabling from 250 MHz to 500 MHz Channel and Permanent Link guidelines and test limits Alien Crosstalk coupling between adjacent 4-pair Cat 6 cabling channels Measured by PS ANEXT and PS AACR-F Additional guidelines for field test equipment and field test methods Accuracy Level IIIe for field certification Alien Crosstalk testing This document is in the final revision stage with final committee ballot Will be published October 2006 14
Channel Length Limitation for Cat 6 / Class E TIA: TSB-155 (Telecommunications Systems Bulletin 155) Support for 10GBASE-T over limited channel length should operate for channel length 37 m should operate for channel length of 37 to 55 m, depending on AXTalk test results may operate over channels > 55 m after mitigation ISO/IEC: TR24750 (Technical Report) Similar limitation for Class E as outlined in TSB-155 for Cat 6 Note: To test is to know: The Alien Crosstalk performance should be (sample-) tested for all cabling links 15 TIA/EIA-568-B.2-10 Draft 4.0 Transmission Performance Specifications for 4-pair 100 Ω Augmented Category 6 Cabling Performance Specifications for all test parameters Worst pair-to-pair and multi-disturber (power sum) requirements for cable NEXT Loss and ELFEXT Alien NEXT (PS ANEXT) and Alien FEXT (PS AACR-F) Performance requirements for Cat 6A components Cable specifications Connecting hardware specifications Critical topic to provide interoperability of Cat 6A Will require significant amount of work past the publication of IEEE 802.3an and TIA TSB155 Patch cord requirements: NEXT and Return Loss General agreement on channel specifications 16
Alien Crosstalk Measurement Method 17 Alien NEXT (ANEXT) Measurement Channel under test Main Disturbed (victim) link Disturber link Remote Synchronization connection Termination Plugs 100 Ω Diff and 50 Ω Common mode Requires a synchronization communication module in the Main and Remote test units. 18
Alien NEXT Measurement End 1 End 2 19 ANEXT Measurements Between Two Links The brown pair is disturbed by the four wire-pairs in the disturber cable The measurement of the Power Sum ANEXT for this disturbed pair requires four pair-to-pair NEXT measurements The PS ANEXT for all wire-pairs requires 16 pair-to-pair measurements Victim Link Disturbers Measure all wire-pair combinations between two adjacent cables in approximately 25 seconds 20
21 Alien FEXT Measurement Main Link under test Remote Synchronization connection Synchronization channel 22
Alien FEXT Measurement End 1 End 2 23 AFEXT Measurement Alternate Main Link under test Victim Link Disturber Link Synchronization connection Remote When testing a horizontal link and a synchronization channel is not available Use the two links under test to synchronize Use the communication module to terminate 24
Alien Crosstalk Pass/Fail Determination for Cat 6A or Class E A PS ANEXT and PS AACR-F for each wire must pass the limit The average PS ANEXT of the four wire pairs must pass the limit established for the Average-Pair The average PS AACR-F of the four wire pairs must pass the limit established for the Average-Pair Pair-to-Pair limit line Average-Pair limit line 25 Alien Crosstalk Pass/Fail Determination for 10GBASE-T Criteria established in previous slide, OR Alien Crosstalk Margin Calculation (ACMC) an alternate evaluation Calculate PSAXTALK for each wire pair Equals the sum of PS ANEXT and PS AACR-F Calculate a PSAXTALK limit line for each wire pair Combined limit for PS ANEXT and PS AACR-F Calculate the average margin for each wire pair over the 10 MHz to 400 MHz frequency range All average margin values must be greater than zero Rationale: The electronics cannot distinguish the effect of ANEXT and AFEXT The victim link is affected by the combination of the two types of disturbances 26
PS AACR-F fails the Pair-to-Pair and the Avg-Pair test The Overall Test results shows Pass based on the outcome of the Alien Crosstalk Margin Calculation (ACMC) 27 The 10 GbE Cabling Test Solution The DTX-1800 Supports frequency range and accuracy ANEXT measurement floor greater than 95 db over the frequency range (1 through 500 MHz) DTX-10GKIT Contains all accessories required to measure PS AXTALK DTX-AXTLK1 DTX-PLA002S DTX-CHA001AS DTX-AXTERM CIQ-RJA DTX-10GCSE AXTalk Communication Modules (2) DTX AxTALK Analyzer Software Cat 6A/Class E A Permanent Link Adapters (2) DTX Channel Adapters for Alien Crosstalk (2) Link Terminator Diff. and Common Mode (2) RJ45-to-RJ45 Adapters (2) Carrying case for items in DTX-10GKIT 28
Alien Crosstalk Measurement Strategy The approach (procedure) to certify a cabling installation for 10 GbE 29 Measurement Strategy Measure the in-channel parameters for every link Frequency range 1 through 500 MHz Upload the results to LinkWare Setup the laptop with the LinkWare test results data and the Alien Crosstalk Measurement Application software Insertion Loss test results must be available in PS ANEXT and PS AACR-F evaluations PS ANEXT and PS AACR-F certification will be based on a sampling method Experiments have shown that PS AXTALK between bundles is negligible Therefore, PS AXTALK testing applies to cables within the same bundle Test the alien crosstalk contribution of all links in a bundle by using all links in the bundle as disturber links 30
Total Possible Number of Tests for PS ANEXT A complete test is not practical ANEXT between two cabling links consists of 16 pair-to-pair combinations plus Power Sum calculation In a bundle of 24 Links: the Power Sum ANEXT calculation for the pairs in one victim link Requires ANEXT measurements of 368 combinations Time estimate: ~17¼ minutes The Power Sum ANEXT measurement combinations when every link must be considered a victim link Requires ANEXT link-to-link measurements of 276 link combinations {4,416 pair-to-pair combinations} Time estimate: 207 minutes or 3 hours and 45 minutes 31 Practical PS AXTalk Strategy 10GBASE-T Specified in IEEE 802.3an, TIA TSB-155 or ISO TR 24570 Select the longest links as victim links The limits are far more demanding for longer links than for shorter links 10GBASE-T only specifies the channel configuration Per TIA TSB-155, you can test the permanent link or the channel configuration. The limits are identical Limited number of links that must be certified for 10GBASE-T Select all links in the bundle of the victim link as disturbers If the links run in the same bundle, test 10% of the bundle Bundle size 12 Test the longest link (one) in the bundle Bundle size 24 Test the two longest links in the bundle Certify a complete cabling plant Select 1% of the links or 5 links whichever is greater 32
Practical PS AXTalk Strategy Cat 6A or Class E A Testing for Augmented Cat 6 (Cat 6A) or Augmented Class E (Class E A ) specifications Select the longest links and some links with the shortest distances between connectors These configurations tend to provide the worst results You have the option to select the channel or permanent link configuration The Pass/Fail limits in the current draft standard are likely to be adapted to special case conditions Note: if you test the channel configuration, all user patch cords must be and must remain in place not be changed! 33 34
Recommendations for Design of New (Data Center) Cabling Systems The 10GBASE-T definition is complete and final Re-test the data center cabling system to certify compliance with the requirements of 10GBASE-T (or TIA TSB-155/ ISO TR 24750) For Channels up to 55 m, Cat 6 UTP offers high probability to meet the PS AXTALK specifications Cabling Decisions The specifications for Augmented Cat 6 (Cat 6A) are not yet complete and some changes are likely Many Cat 6A cables have a larger OD to put distance between wire pairs in adjacent cables Impact on pathways: fill rate and size Impact on wire management: larger bend radius 35 Recommendations (Cont d) Keep bundles small (12 links may be an optimum size) Alien XTALK of wire-pairs in different bundles is most often negligible Smaller bundles mean fewer links need to be tested for Alien XTALK interaction Space the bundle ties to 2 or 3 feet; do not over-tighten less rigid bundles create less AXTALK Incorporate the bundle or pathway name in the Cable ID If you need to test cabling links for AXTALK performance, you only need to test among links in the same bundle 36
Recommendations (Cont d) Select a quality installation contractor High quality workmanship will preserve Cat 6 or Cat 6A component specifications Test and Certify the cabling to one of the new standards (latest draft if not yet ratified) TIA TSB-155 or TIA-568-B.2-10 ISO TR 24750 or ISO 11801 2 nd Amendment Testing is the only way to know that the completed installation meets the requirements and offers some safety margin 37 Summary / Conclusions Cable certification for 10 GbE requires two steps: 1. Test the cabling parameters (In-Channel) from 1 through 500 MHz Use the DTX-1800 as is and select appropriate standard Store plot data and upload to LinkWare Organize test results data by bundle 2. Test the Alien Crosstalk parameters for selected cabling links Use DTX-1800 with DTX-10GKIT Selected victim links should be longest (per 10GBASE-T) Include all links in the same bundle as disturbers Save time and effort by Keeping bundles small (if possible less than 24 link) Including documentation of bundle/pathway in Cable ID 38
Time for questions Contact Information: Eric.Lie@flukenetworks.com 39