Testing challenges of 40/100Gbit ethernet network infrastructure

Migration to 40/100Gbit Ethernet is presenting the telecommunications industry with a way of achieving the bandwidth levels now being expected by data hungry consumers, while at the same time benefiting from the added functionality and enhanced flexibility of an all-IP network architecture. However, this advanced optical networking technology will also require a more sophisticated approach testing, so that communications infrastructure using it is fully effective. This is due to the fact that client interfaces are based on multi-wavelength parallel optical transmission, rather than the conventional serial transmission employed in current network infrastructure.

The motivation for upgrading to higher speed optical network technology is fairly straight forward. Firstly there is a need to augment the mobile backhaul in order to support the growing bandwidth demands of 3G/LTE communication. Secondly there has been an enormous increase in general bandwidth requirements across wire-line networks in recent times, with an ever growing number of Internet users along with a greater variety of applications becoming popular.

Implementation of100Gbit is likely to prove difficult as this is potentially a very disruptive technology, especially at the physical layer with data being transported over 4 x 25Gbit/s (or 10 x 10Gbit/s) optical wavelengths arranged in parallel. This allows the bandwidth capacity to be raised by an order of magnitude, carrying hundreds of millions of data packets per second. Those 4 or 10 wavelengths are converted into 10 electrical lanes each with 2 x 5Gbit virtual lanes. This raises the complexity level of each implementation – brings new errors and alarms at the physical level, and necessitating a whole new, far more sophisticated test approach.

Currently network equipment manufacturers are utilising C Form-factor Pluggable (CFP) transponder modules of two different types. The physical form factor has been specified by the CFP MSA group. The 4 x 25GBit/s LR4 interface type (100GBASE-LR4) as standardised by the IEEE in 802.3ba will likely become the dominant type.

Major considerations when testing next generation optical networks include:

Carrying out even the simplest of lab based tests, such as checking the optical power of a 100Gbit interface, will not be possible with traditional power meters. This is because such devices aggregate the power of all wavelengths. Therefore use of optical channel checkers or optical spectrum analysers will be needed to measure the optical power/wavelength

To ensure interoperability between different CFP modules, checking standard compliance and operating margins before moving into higher layer testing is recommended.

With the concept of parallel lanes, skew testing (layer 1) is of prime importance. Compliance to skew specifications can be a major issue effecting 40/100Gbit Ethernet implementations, with signals arriving at slightly different times. Some constituents of skew are static in nature – this is caused by data being transported on multiple lanes each of which might have slightly different parameters and different time of flight values. These remain the same over time. The main addition skew element is dynamic skew – this is primarily caused by differences in the characteristics of optical components in the network (due to the influence of factors such as temperature). Tolerance parameters for both static and dynamic skew are defined by IEEE and should be stress tested prior to deployment to ensure error free operation when customer traffic is passing through the link.

Bit Error Rate (BER) tests need to be conducted on a per lane basis to be able to find the root cause of bit errors associated to physical errors more easily (e.g. if they only appear on specific lanes or if they are spread out over all lanes). Once the testing of these new parallel implementations has been successfully conducted, it is then also recommended to run conventional Layer 2 /3 Ethernet IP tests, using RFC-2544 or Quality of Service (QoS) testing on multiple streams of traffic. This ensures proper traffic prioritisation within the network.

The equipment needed to carry out the tests described represents a significant investment. Given the current economic climate, operators need to maximise the revenue they generate from new high speed network infrastructure while keeping the spend on costly test hardware as low as possible. Rental of equipment is now being seen as a highly attractive way of achieving this. It allows operators and their contractors to avoid having to make capital investments, as well as being able to negate the various ongoing costs that come with ownership, – namely maintenance/upkeep, downtime cover, recalibration security and insurance. By taking a rental approach, test equipment can be expensed as operational cost and while it is in regular use, can thereby generate more than enough revenue to justify itself. Furthermore, network equipment manufacturers will also begin to consider renting test solutions as a viable alternative, such that they can show any new client they are working on a 100Gbit/s project leading to trials. This means the manufacturer does not have to commit to purchasing test items early on, leaving the option to purchase after the trials are completed and the project is committed.

The rising popularity of bandwidth-intensive services, with subscribers downloading, streaming and sharing of multimedia content, as well as the growing prevalence of online gaming and video calls/conferencing has huge implications for the future of the telecoms industry. The greater importance of the Internet in our daily lives and an ongoing increase in the number of global users is forcing carriers to upgrade to higher speed networks with high density hardware.

Test solutions for 40/10Gbit Ethernet transmission enable operators to carry out all required tests ranging from the physical layer to the IP layer. Utilising test equipment with tighter specifications and having a greater breadth of test metrics will enable carriers to deploy high speed network infrastructure more efficiently and thus fully capitalise on commercial opportunities it presents.

The roll out of 40/10Gbit Ethernet networks will not only result in a change in how testing procedures are undertaken, but also call for a re-evaluation of how carriers gain access to test equipment. Due to the costs involved, more forward-thinking carriers and their contractors are now starting to change their sourcing strategies. This has led test equipment manufacturer JDSU and sourcing specialist Livingston to work together so that their customers can find the means of obtaining state of the art test hardware that best suits their specific financial circumstances at that time and gives them the highest degree of flexibility.

Authors: Geoff Kempster, Livingston & Olaf Herr, JDSU

Check Also

Can extended warranty ever equal reliability?

Does the increasingly common practice of offering an extended warranty with a new EMC or …