New Revenue Opportunities Creates Testing Challenges
The insatiable need for more bandwidth and the constant development of high-bandwidth IP and data applications for today’s consumers and businesses is driving service providers, and the network equipment manufacturers (NEMs) that supply them, to find quick routes to deploy Ethernet at 40 and 100 Gigabits per second. Originally, it seemed that service providers were going directly to 100 Gigabit Ethernet (100GE). But after considering the cost and the complexity of 100GE optics, some decided to start with 40 Gigabit Ethernet (40GE) instead. 40GE will also play an important role in aggregation within data centers as it should ride a more aggressive price and port power/density curve.
These developments present some significant optical and electronic testing challenges for the teams that will be responsible for validating and maintaining the new equipment. For example, 40GE and 100GE are transported over 4 or 10 separate electrical lanes respectively so the relative skew between lanes needs to be characterized and stressed. Jitter and wander measurements are common in the telecom world for synchronous networks but Stressed Receiver Sensitivity (SRS) tests have been specified instead for Ethernet networks, especially at higher speeds. The ratification of the SRS standard for 100GE is expected for early 2010. Service technicians will need, and suppliers are preparing to deliver, test and measurement solutions that will meet the needs of the installation, turn-up, and operations phases.
Super High Speed Lanes
A few key users such as server farms, cloud computing vendors, large Internet nodes, and large research labs are driving today’s quest for ever higher connection and aggregation speeds. Up to now, super services including Google, Facebook, Yahoo, and others have been forced to use multiple 10GE links in parallel to meet their bandwidth needs but this presents problems with port density, traffic management, and power. Now they are desperately looking for higher capacity. These super services are looking at both 40GE and 100GE for switches and routers and if they choose 100GE this will drive 100GE into the transport ecosystem. Service providers competing for the business of such customers are rushing to bring products to market quickly.(See Figure 1.)
The Institute of Electric and Electronic Engineers (IEEE) is working on the 40GE and 100GE standard (IEE 802.3ba) which is expected to be ratified in 2010. The current 40GE draft defines 4 lambdas running at 10Gbps each while 100GE is based on four 25Gbps lambdas. On the electrical interface side, both 40GE and 100GE will utilize respectively 4 and 10 parallel electrical 10Gbps lanes, each carrying embedded coding information.
Network equipment manufacturers spent the past year primarily working the development phase for 40GE/100GE offerings. The CFP Multi-Source Agreement (MSA) is the first industry standard to support next-generation Ethernet optical transceivers for both 100GE and 40GE. The CFP MSA defines the form factor of a hot-swappable pluggable optical module supporting 40GE or 100GE using an electrical interface consisting of multiple 10Gbps lanes.
The CFP MSA is supported by industry heavyweights such as Finisar, Opnext, and Sumitomo/Excelight. Current CFP modules are designed to support a reach of up to 40 km. The first CFP pluggable form factor optics for 40GE/100GE became available at the end of 2009. CFP modules will also support the Optical Transport Networks (OTN) system, running at 112G. The OTU4 transport layer adds the framing, management layer and forward error correction on top of Ethernet that is needed for longer distance transmission and for OTN management in a carrier network. The OIF is working on an MSA agreement for guidelines for DWDM transport optics for 40GE and 100GE. This should lead to a range of modules with defined standard mechanics and electrical interfaces for long haul of 40GE and 100GE some time in 2011.
First-Gen 40GE/100GE Products
The first generation of 40GE/100GE products will likely be data-centric devices utilizing new Transmitter Optical Subassemblies (TOSA) and Receiver Optical Subassemblies (ROSA) arrays, optical mux/demuxes and high speed ICs operating at 25Gbps. Long haul transport, on the other hand, will require a new generation of optics including much more complex TX and RX using novel modulation schemes, coherent detectors and complex, DSP based post processing schemes. As a result, long haul optical systems for 100GE will take longer to reach the market.
A few key differences help explain why 40GE is expected to be considerably easier and less expensive to implement than 100GE. First of all, 40GE uses familiar 10Gbps optics technology while 100GE uses 25Gbps optics that are new and currently present some interesting technical challenges. Secondly, the mismatch between the 25Gbps optical layer and the 10Gbps electrical layer in 100GE modules requires resolution. The physical coding sublayer is responsible for bonding multiple lanes together through striping or fragmentation techniques.
The 100GE module uses 20 virtual 5Gbps PCS lanes to connect 10 x 10Gbps electrical lanes to 4 x 25Gbps optical lanes. This operation is usually implemented in a high speed IC, that’s considered to be the "gearbox". This gearbox is used to map individual 10Gbps lanes onto the four 25Gbps lambdas. The gearbox represents a major technical challenge which places limits on power, cost, and port density. The long-term evolution will be towards native 25G lanes in both the electrical and optical domains, providing a clean roadmap to 400GE. 40GE will offer advantages in cost, port density, and power but 100G will also experience growth in critical high bandwidth applications.
The high-price premium of early 100GE optics may be a barrier for initial deployment of 100G. Some providers may use 40GE technology as a steppingstone, allowing them to scale up to 100GE at a later date. 40GE and 100GE will also likely be mixed to match the varying bandwidth requirements in different parts of the network. Equipment manufacturers will begin to offer interfaces for both 40GE and 100GE early in 2010 so users will be able to choose based on their bandwidth and density requirements.
Testing the Pit Crew (Service Team)
The service team is tasked with addressing the testing challenges posed by the new 40GE/100GE technology. For example, service technicians are used to working with multiple lambdas in wavelength division multiplexing (WDM). But the multiple lambdas used in WDM transmit independently while 40GE and 100GE lambdas are part of a single data stream. As a result, an interruption of one lambda in a system of bonded 10GE links will only decrease the capacity of the bonded link by 10Gbps. On the other hand, an interruption of a lambda of a 100GE system will break the whole 100GE link. (See Figure 2.)
40GE and 100GE systems are designed for short distances up to 40km. First generation systems will be designed for a 10km reach. The 4 lambda used in 100GE are between 1295nm and 1310nm with a 5nm spacing. This means that the optical Signal to Noise Ratio (OSNR) will not be critical. However, the optical power levels should be measured to ensure the power budget and the correct wavelength must be verified to match the optical filters in the optical module demultiplexer.
Chromatic dispersion (CD), polarization mode dispersion (PMD), optical return loss (ORL), and attenuation are essential measurements for fiber and link characterization prior to installation of 40GE and 100GE long haul transmission systems. The combination of CD, PMD, ORL, and attenuation test functions in a single instrument allows technicians to validate the fiber link’s compatibility with high-speed WDM systems.
The test platform delivers multiple optical signals and then applies stress to the signal such as modifying the eye closure. Electric lane skew can be added to optical lane stress to fully test the margins of the electro-optics. All of the current optical measurement benchmarks such as power, stability, SRS, and eye-pattern will still be important but they will have to be met over 4 lambdas.
As communications providers upgrade their networks to 40GE/100GE, they must provide customers with extremely high service quality levels. Closely coupled to this goal is the ability to measure network performance either proactively or in response to trouble tickets. The latest generation of test solutions addresses the full range of service providers’ 40GE/100GE testing requirements while also meeting optical transport network (OTN) and synchronous optical network/synchronous digital hierarchy (SONET/SDH) test functionality in one platform. The new tools enable service providers to deploy, commission, and maintain these new services quickly, efficiently, and conveniently. Service technicians will not be required to carry and maintain multiple test sets in support of today’s advanced service offerings.
Reading the Right Signals
Service providers may also need to perform electrical signal testing of 40GE and 100GE as part of the equipment or optical module selection and troubleshooting process. 100GE electrical testing includes measuring static and dynamic skew of multilane signals at the full rate to verify the functionality of the equipment. If the time delay between signals on different lambdas (skew) goes beyond a certain value then the data in the buffer overruns and data gets lost. This means that excessive skew between lambdas will generate bit errors. Generating static and dynamic skew will help to assess the margin of the equipment to skew effects.
The mapping of individual lanes onto lambdas is random at startup so you normally cannot generate an aggregated (per 25G lambda) BER to determine the full performance vital in the development and evaluation phase of CFP modules. The ability to group the individual lanes on a per-lambda basis greatly increases the depth and breadth of testing that can be done over the link. This ability quickly allows operators to identify and diagnose errors in the optical domain (poor optical performance) or electrical layer (crosstalk or timing).
The optical module and the line card can be tested by applying framed and unframed signals and adding payloads to framed signals to verify error free operation. A pseudo-random bit sequence (PRBS) or digital word can be used to stress both the optical and photonic layers to identify faults that are impossible to troubleshoot at higher layers. Test equipment offering fast, low jitter triggers can work together with fast oscilloscopes to reconcile the physical and logical signals.
PCS and IP Layer Testing
The next step is typically to stimulate the system with PCS data while testing the optical module and the line card. Major considerations on the receiver side include mapping the received to the transmitted PCS lanes and generating and measuring virtual lane skew. The PCS layer alarms and errors need to be fully validated against standards. Test applications that allow full control of the coding and payload will help in troubleshooting the PCS layer. The IP layer can then be tested by sending and receiving IP frames. A test instrument that can create a full 40GE/100GE load on the client either as a single flow or in multiple flows is ideally suited for this task. (See Figure 3.)
A new generation of electrical layer test solutions helps service providers, component developers and network equipment designers unlock the performance and revenue potential of 40GE/100GE technology. The full range of solutions provides testing of:
1. Optical and electrical interfaces from the physical layer to the Ethernet/IP, protocol testing, PCS layer validation and optical module testing.
2. Stressed receiver sensitivity on 40GE/100GE systems including stress generation and stress sensitivity measurements.
3. Multiplexing de-multiplexing, signal conditions and signal access of 40GE/100GE optical signals. This range of testers covers the full electrical-layer requirements for 40GE/100GE including optical module testing/validation, network equipment development and system verification testing. Advanced applications for the PCS and the IP layer support the challenging task of debugging and verifying complex 40GE/100GE multi-lane, multi-lambda products.
OTU4 will play an important role in future network equipment, and not just in the core transport network. The ability to monitor the performance of both the FEC (forward error correction) and overhead bytes gives a detailed view of the health and condition of a link. It is important that equipment be ready to test these OTN client ports that will play an increasing role in and out of the core network.
The Route to Success
A complete suite of testing solutions gives service providers the confidence to successfully unlock the performance and revenue potential of 40GE/100GE technology. These solutions enable service providers to adopt and install 40GE, 100GE, and OTU4 equipment at critical aggregation points to ensure full and seamless validation and verification of interoperability with current equipment and OTU4. The net result is that service providers gain control and certainty that helps them bring high revenue earning 40GE/100GE services to market faster and more reliably.