Cell Backhaul: Realities of Ethernet in SLA Environments Presented by Zach Sherman Applications Engineer Transition Networks MBA, BSEE

Agenda Situational Analysis Upcoming Bandwidth Reality Choices to make Can Ethernet perform like TDM Ethernet Standards to know

Situational Analysis 200 Trillion Text messages are received each day in America An increase by almost 500% in two years 5 Billion Applications were downloaded in 2010 Mobile web users up more than 33% Over 50% of phones sold in 2011 are “smart” phones (<25% in 2009)

Upcoming Reality Not enough Bandwidth to towers Each iPhone/Droid user Consumes >560M per month Each tablet user Consumes >800M per month That’s a lot of T1’s

Choices to Make Do I put more T1’s to towers? Bandwidth growth has been exponential How many more T1’s will be enough? For two years? For five years? Do I put Ethernet to towers? How? The infrastructure won’t support it Only copper to the tower If I have the fiber how do I get Ethernet to work QoS SLA’s Dropped Calls

Ethernet’s Reality Ethernet might provide the bandwidth but can I get the service I expect from T1’s? Traditional Ethernet was non-deterministic Half-Duplex CSMA/CD Random backoff delays Destruction and Re-transmission of data Switched Full-duplex Ethernet Removes CSMA/CD Each port is its own collision domain in Full-duplex 802.1p/Q VLAN’s and Prioritization Real-Time Ethernet The Foundation of Many SLA’s is Time

Ethernet Technologies Aimed at SLA’s VLAN Tagging Q-in-Q Traffic Classification Quality of Service Techniques Class of Service (CoS) Differentiated Services (DiffServ) IEEE 802.3AH (Link OAM) IEEE 802.1AG (Service OAM) ITU-T Y.1731 (Performance Monitoring) Synchronous Ethernet (SyncE) IEEE 1588v2 G.8032 Ring Protection

IEEE 802.1Q and VLANs IEEE 802.1Q, (dot1q) allows logical network connections to share the same physical network VLAN’s logically separate broadcast domains at layer 2 A VLAN is typically used to isolate traffic types logically in an organization while sharing the same physical network Both sales and engineering need access to the network, but each has its own VLAN so that information is protected from one segment to another The VLAN tag is a two-byte (16 bit) frame used to identify the traffic circulating on the VLAN Contains a three-bit user priority (CoS tag) One-bit canonical indicator 12 bit VLAN ID

802.1ad – Q-in-Q Q-in-Q commonly referred to as double tagging Allows tagged traffic to be preserved while adding additional tags Useful for Internet service providers, allowing them to use VLANs internally while mixing traffic from clients that is already VLAN-tagged. Customers can use and manage their own C-VLANs for each user group whereas Service Providers can use their own S-VLAN to isolate traffic from each customer into a VLAN An Ethernet frame with Q-in-Q looks like a VLAN-tagged frame, except that it has two tags instead of one DA SA VID 20 Type 88A8 30 8100 Data FCS 802.1ad 802.1Q

5-VLAN Tagging Types

QoS Techniques

Why Quality of Service? Delay Sensitive ● High Tolerance For Delay Voice - Internet Browsing Streaming Video - Email Video Conferencing - File Transfer QoS defines rules for processing packets Based on priority or weight Class of Service (CoS) Type of Service (ToS)

IEEE 802.1P Class of Service Class of Service (CoS) 3-bit value Commonly Called the P-Bit 3-bit value Typically associated with a VLAN ID Value of 0 to 7 7 being the highest priority traffic Doesn’t Mean Much without Rules

DiffServ – IP ToS IP Value Priority Type of Service – ToS Replaced by DiffServ Uses 6 bits in the IP header Allows for 64 traffic classifications Table of the Most Common Types Doesn’t mean much without processing rules IP Value Priority 101 110 High Critical 000 000 Best Effort 001 010 Low 000 100 Medium 001 110 High

Two Types of QoS Hard QoS Soft QoS Not a dedicated amount of bandwidth Reserves a selected amount of bandwidth for a traffic type No other traffic types can use this bandwidth Soft QoS Not a dedicated amount of bandwidth Allows for flexibility in assignment and re-assignment

Rule Systems Strict Priority Weighted Fair Queuing Allows an administrator to determine exactly how much bandwidth is allowable to each flow More difficult to implement because of the interface to control Weighted Fair Queuing Weighted Round Robin Queuing Allows for certain traffic to get a automatically assigned bandwidth percentage Automatic nature is simple to implement

Flow(1) +Flow(2) + Flow (3)…+ Flow(n) WFQ Example Based on the inbound CoS or ToS tag, each traffic type is assigned a weight for processing. This weight determines bandwidth percentage Priority Queue Frame Type Weights Best Effort 1 Background 2 Excellent 4 3 Critical 8 WFQ equation: B*Flow (1) Flow(1) +Flow(2) + Flow (3)…+ Flow(n)

802.3ah Link OAM AKA – Ethernet in the First/Last Mile Provides for IP-less management of remote nodes between vendors Cisco talks to HP, etc Provides critical link fault information Last Gasp/Dying Gasp Link Failure Critical Event Is Point to Point only (Direct Connection)

Customer End-to-End Metrics Provider End-to-End Metrics 802.1AG/Y.1731 Introduction Service Provider Operator Customer End-to-End Metrics Provider End-to-End Metrics Operator Metrics As Ethernet continues to replace legacy TDM services in QoS sensitive, high-capacity applications such as business services and wireless backhaul, ensuring service quality meets customer expectations requires a well managed, operationally efficient network. Ethernet connectivity and service layer Operations, Administration and Maintenance (OAM) standards are designed to simplify the management of Carrier Ethernet services with end-to-end service visibility, fault isolation, reporting and continuous performance monitoring. As specified in the IEEE 802.1ag and Y.1731 standards, these capabilities enable providers to manage Ethernet services regardless of the network path, topology, operators or network layer that carries the traffic between service endpoints. Connectivity Fault Management (CFM), defined in both 802.1ag and Y.1731, divides the provider’s end-to-end network into three distinct levels or maintenance domains: customer, provider, and operator (e.g. a partner carrier’s network). CFM respects this hierarchy by ensuring that faults identified in a lower layer (e.g. operator’s network) are alarmed only to the next higher level (e.g. the service provider) so that appropriate action such as traffic rerouting can be performed, while the details of the problem (fault isolation) remain at the layer where the incident occurred. This ensures that the fault is regulated within the appropriate domain (by the operator or provider), while preventing a mass broadcast of alarms throughout all layers of the network. The key features of Ethernet OAM CFM are Fault Detection, Verification, Isolation and Notification that will be covered in the following slides. Some other terms used in 802.1ag that you should become familiar with are: Maintenance Association (MA) – Boundaries of an Administrator’s scope of monitoring part of the network Maintenance Domain (MD) – A level of monitoring within the hierarchy Maintenance End Points (MEP) – End Points of the MA or MD Maintenance Intermediate Points (MIP) – Intermediate Points within MA or MD www.transition.com

Continuity Check Messages Fault Detection Connectivity Check Messages (CCMs) are periodic messages used for detecting loss of continuity within an MA Each MEP transmits CCMs to all other MEPs in the MA Upon loss of 3 consecutive CCMs a loss of continuity defect is declared NOC CCM Timeout Alarms EVC Failure CCM Alarm

Loopback Fault Verification Works with central test head to perform tests Measures performance (delay, dropped packets, throughput, etc.) Ideal for fault isolation and locating within carrier network Port level loopbacks are ideal for turn up and commissioning Eliminates truck rolls NOC EVC Failure Loopbacks can be used for Fault Verification. Ethernet loopbacks using 802.1ag can be thought of as the equivalent to the IP “Ping” command. Service faults can be verified using a Loopback Messages (LBM) and Loopback Replies (LBR). A series of Loopback Messages (LBM) can be sent to identify the location of the fault by querying maintenance endpoints (MEPs) and maintenance intermediate points (MIPs) along the service path. Loopbacks are also very effective during service turn-up and commissioning. You can put the far end device into a loopback and run tests from a head-end location to measure a variety of performance statistics for the circuit. These performance items may include items such as delay, dropped packets and throughput when commissioning an Ethernet circuit. 802.1ag loopbacks do not happen automatically and they require the intervention / activation of someone from the Network Operations Center. LBM LBR www.transition.com

Linktrace Fault Isolation Quickly determine the exact location of a fault Tracks the entire path Hop-by-hop Similar to IP Trace Route function NOC EVC The location of a fault can be quickly determined by a Linktrace Message (LTM). The Linktrace function is analogous to the IP trace route function. When a Linktrace Message (LTM) is sent to a maintenance endpoint (MEP), all maintenance intermediate points (MIPs) respond with a Linktrace Reply (LTR) along path traveled by the Linktrace Message (LTM). The returned Linktrace Replies (LTRs) (and those not returned) uniquely identify the segment or node where the fault originates. Under normal operating conditions, Linktrace is also used by network elements to determine the path a service takes through the network – this route awareness is stored in a local database to expedite fault isolation, and for link protection purposes. 802.1ag linktraces do not happen automatically and they require the intervention / activation of someone from the Network Operations Center. Broken Link LTM LTR www.transition.com

AG/Y.1731 OAM Summary  OAM Function 802.1ag Y.1731 Method CFM Fault Detection  CCM Fault Verification / Loopback LBM / LBR (“Ping”) Fault Isolation LTM / LTR Discovery LTM / LTR & Multicast LBM* Fault Notification AIS / RDI PM Frame Loss CCM, LTM / LTR Frame Delay DM (1 way), DMM / DMR Delay Variation In another presentation, we already discussed the features and benefits of IEEE 802.1ag Service OAM. This slide shows the primary differences between IEEE 802.1ag and ITU-T Y.1731 in a succinct check list. You can see that 1731 covers everything that AG does as well as adding some features. One of those features falls under connectivity fault management or CFM on the chart. That feature is a specific provision enabling fault notification through use of 1731 messaging. While this is a minor contribution to AG’s functionality, where Y.1731 separates itself is in Performance monitoring or PM on the chart. Y.1731 offers Frame loss, frame delay and delay variation to help implementers monitor per-flow traffic statistics. www.transition.com 24

Packet Statistics Y.1731 Provides Transmission Stats Frame Delay (FD) Delay Measurement Messages (DMMs) Delay Measurement Responses (DMRs) Frame Delay Variance (FDV) The maximum FD less the minimum FD FD is an average measurement of delay Frame Loss (FL) Continuity Check Messages (CCMs) Link Trace Messages (LTMs) Link Trace Responses (LTRs) Frame Loss Ratio (FLR) Percentage of frames reaching destination

Synchronous Ethernet SyncE ITU recommendation G.8261 The third Ethernet clocking method in this presentation is commonely referred to as Synchronous Ethernet or SynchE. It is based on ITU Recommendation G.8261 and essentially creates a method that the Primary Refernce clock singal enters the bit stream similar to SONET networks and every node in the packet network recovers the clock from this stream. One primary advantage that SynchE creates is that the clocking is entirely independent of the network loading making it reliable and accurate in highly congested networks.

SynchE ITU recommendation G.8261 Uses the Physical Layer of Ethernet Clock Singaling is kept separate from Data Traffic High reliability and accuracy A Primary Reference clock is inserted through a separate clock port Does not interfere with existing IEEE protocols Uses OAMPDU’s for delivering Synchrozation Messages (SSMs) Used only for synchronization of clocks Does not distribute Time of Day (ToD) messages

IEEE 1588v2 Precision Timing Protocol (PTP) Independent of the Physical Layer Uses Packets to transport timing information Sends Time of Day (ToD) and Synchronization info Can be affected by network delays and jitter Can be used in conjunction with SyncE SyncE delivers accurate Frequency/Sync information 1588v2 delivers ToD

G.8032 Ring Protection Ethernet Ring Protection Switching (ERPS) Sub 50ms Recovery Protects Ethernet Rings from Link and Node failures Offers Ethernet SONET/SDH ring failover and deterministic performance

Three Solutions from TN 32xT1 over Ethernet 802.1ag/Y.1731 NIDs SyncE, 1588v2, G.8032 NIDs

32xT1 over Ethernet PacketBand-TDM Line

T1’s over the Ethernet Core

Keys to T1 over Ethernet Clocking Ethernet Processing PacketBand exceeds the G.824 SyncMask Standard for Clock Recovery <16ppb clocking error rates Ethernet Processing On Board Ethernet switch Support VLAN’s, QoS (CoS, DSCP/DiffServ) Additional Ethernet ports, fiber (SFP & Cu) supports Ethernet and TDM delivery ‘in the box’ rate limiting egress queue prioritization

802.1AG/Y.1731 NIDs x3230 Single Port NID with Failover Fully supports: VLANs Q-in-Q CoS (802.1P) DiffServ Strict Priority Queuing 802.1AG Y.1731 Sub 50ms fiber failover

802.1AG/Y.1731 NIDs S3240 Multi-Port NID with Failover Fully supports: VLANs Q-in-Q CoS (802.1P) DiffServ Strict Priority Queuing 802.1AG Y.1731 Sub 50ms fiber failover Dual Redundant DC Power inputs

Uber-NIDs S3280 & SISGM1040-384-LRT Are the basis for all future NIDs Adds features to S3240 More ports (8 total) 2 Out of Band Management ports IPv6 1588v2 SyncE G.8032 (-40) to 75C temp range IEC61850 (on SISGM) Are the basis for all future NIDs Includes plans for 10G

Summary Many challenges related to increasing bandwidth requirements for mobile services x number of T1’s may no longer be enough Ethernet is a viable option for bandwidth Groups like the ITU, MEF, and IEEE are all working on continuously improving Ethernet to provide the type of reliability and availability inherent to TDM circuits Transition Networks is active in these groups and is staying as informed on emerging standards as they are ratified by the various groups