How will Cat 8 redefine the limits of copper? - Ravi Doddavaram Psiber Data
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
How will Cat 8 redefine the limits of copper? Ravi Doddavaram Psiber Data rdoddavaram@psiber-data.com
Traditional Datacenter Topology Still Not Interesting 40 Gbps (LOM) or LAG 10 Gbps(LOM) Romely for instance Virtual NICs push bandwidth requirements on Servers/Access
LAN/SAN Convergence SAN is the biggest driver for bandwidth Most of the adoption for 40GBase Technologies are driven from here FCoE will enable a unified network architecture if Media supports, and no cost penalty
LAN/SAN Convergence SAN is the biggest driver for bandwidth Most of the adoption for 40GBase Technologies are driven from here FCoE will enable a unified network architecture if Media supports, and no cost penalty Virtualization will make the figure on the right look like a big cloud!
“Datacenter Networks are in my way!” What is driving 40G on Copper/Fiber? Change in Data Center Topology Moving away from traditional 3 Tier to 2 Tier (more virtualization) LAN/SAN convergence (Storage Speeds far exceed Network speeds) Growth of 10G Base-T LOM Integration of 10G Cost of Base-T solution is significantly cheaper than SR solution when possible New on board 10G servers (like the Intel “Romley” based servers with 10G MACs) 28nm Technology driving power down (1.5W/10GBase-T port)
40G Pit Falls Adoption will depend on three criteria Power Consumption With 28nm Technology, power ratio is about 2:1 as fiber IEEE has taken this into consideration hence higher bandwidth of cabling (Next slide) Reach objective Not a 90m link, but should support longer length than DAC Advantage of 40GBase-T for supporting 30m reach (EoR, MoR) configuration Cost per Gb and utilization efficiency DAC supports up to 7m but utilization efficiency is poor Active links are expensive LOM of 10G will drive cost per Gb down
Power Consumption Issue: Addressed with higher bandwidth 40G Base-T Class I, 1.6 GHz 40G Base-T Class II, 1.6 GHz Source: ISO11801-99-1
Current status of Standards • IEEE 802.3 NGBASE-T Study Group has become the IEEE P802.3bq 40GBASE-T Task Force. The task force has commenced their work in the creation of a draft standard for 40GBASE-T. Expected by Feb’2016. Work has begun in defining the PHY specification • IEEE is influencing alignment between TIA TR-42.7 and ISO/IEC JTC 1/SC 25/WG 3 on specifications for cabling systems being developed for use with 40GBASE-T • Current draft of TIA 568 C.2 for CAT 8 is 0.9. Defined most parameters for cabling including Frequency, Channel performance • Liaison from IEC TC46/WG9 to IEC SC25/WG3 for 2GHz Field Testing to be sent soon, draft is agreed upon • Field Tester standard, Level 2G task group has been established
TIA Category 8 1 MHz to 2000 MHz Support for 40GBASE-T (IEEE P802.3bq) Insertion Loss (IL) is critical parameter 30m Channel Length, 26m Permanent Link Length Approx. 55% of Channels
TIA Cabling for IEEE 802.3 40GBASE-T (draft) Class I Class II 2000 MHz 2000 MHz New & New & Improved Cat6A Improved Cat7A Components Components 30m 30m 2 Connectors 2 Connectors
TIA Insertion Loss Limit 0 200 400 600 800 1000 1200 1400 1600 1800 2000 0.00 5.00 10.00 15.00 Cat 8* 20.00 25.00 30.00 35.00 40.00 45.00 50.00 55.00 Cat 5e Cat 6 Cat 6A 60.00 *CAT8 limits are TIA Cat 8 CH IL TIA Cat 6A CH IL TIA Cat 6 CH IL TIA Cat5e CH IL based on early draft
ISO Cabling for IEEE 802.3 40GBASE-T (draft) Cat7A Cat7A Class I Class II 1000 MHz 1600 MHz 2000 MHz 2000 MHz Standardized New & New & Standardized Improved Improved Cat7A Cat7A Cat6A Cat7A Components Components Components Components 25m 30m 30m 50m 2 Connectors 2 Connectors 2 Connectors 2 Connectors
ISO Class I / Class II : Insertion Loss Limit 0 200 400 600 800 1000 1200 1400 1600 1800 2000 0 5 10 FA 40G* Class I* 15 20 25 30 35 40 45 50 FA 55 60 65 D E EA F Class II* 70 ISO Class I CH IL ISO Class II CH IL ISO 40G CH IL ISO FA CH IL *limits are based on ISO F CH IL ISO EA CH IL ISO E CH IL ISO D CH IL early draft
Typical CAT 8 Measurements The next few slides present an experiment done in Singapore with a prototype cable that qualifies for the draft ISO specifications in National Metrology Center (NMC) The experiment correlates the measurements done on the cable using a laboratory Network Analyzer in NMC and a field tester Results were published in TIA, ISO and IEEE meetings and extremely well received
Test Setup
Test Adapters & Calibration Artifacts THRU / OPEN / SHORT / LOAD
Measurement Setup with Network Analyzer
Measurement Setup with Field Tester Seminartitel
Return Loss 17.08.2007 Seite 22
Near End Crosstalk
Insertion Loss
Known Issues with high frequency cable designs Insertion Loss suck outs • Typically seen in screened systems above 1300 MHz. • Usually the cause is the improper shield termination to connectors, or shield lay-lengths.
Alien Crosstalk Field Measurements
Alien Crosstalk Measurements UTP Cable ANEXT -40 0 200 400 600 800 1000 1200 1400 1600 1800 2000 -60 "12-12-NA" -80 Pair_12-12-WX -100 -120 F/UTP Cable ANEXT -50 0 200 400 600 800 1000 1200 1400 1600 1800 2000 -60 -70 -80 "36-12 -NA" -90 Pair_36-12-WX -100 -110 -120
Alien Crosstalk: Is it still the white Elephant? Still a very key measurement for 10G Base-T installations. Several installation issues go un-noticed Field testing doesn’t have to be complex anymore One of the few measurements that directly correlates to BER (application bandwidth). TCL, ELCTCL are secondary effects which show up as Alien Crosstalk or Near End Crosstalk Perhaps not so important in the shielded world (Realm of CAT 8) but work still in progress
Additional Measurements? ELTCL, LCTCL, Coupling Attenuation • Measure of balance of the twist of cabling and noise rejection of the cable • Not very meaningful in the field because it is a inherent property of the cable or connector, not much can be done in the field to rectify it • Will be interesting to see how repeatable the measurements are with different common mode impedances of different cables in the field • A fault in ELTCL, Coupling Attenuation will always result in a NEXT, Alien Crosstalk failure. Instead of tightening limit for a new measurement might as well look at the NEXT performance • ISO/IEC liaison letter to TC46 requests the committee the need to evaluate suitability of this measurement in the field along with Coupling attenuation measurement
End to End testing: Concept Existing configuration Additional Test Configuration Channel under Test Field Tester Field Tester Comparison with network analyzer is a problem due to soft de-embedding in channel testing Issue typically seen at higher frequencies Proposal to include “Patch cord” type test configuration for CAT 8 cables Widely supported but no additional effort required as Field testers already support “Patch Cord configuration”
The End Psiber Data is an active member on both TIA & ISO Cabling committees. Level V Field Tester Standard expected soon (All of us field test manufacturers claim compliance, but the standard is still in draft state) Discussion on Level 2G has started (Please note the difference in the name no more level IV, V) Thank you Please email your questions & comments to rdoddavaram@Psiber-data.com
You can also read