The following are notes and thoughts I have taken from an awesome presentation by Richard Steenbergen at Nanog in 2016.

Most of us are typically accustomed to copper mediums when it comes networking specially at the campus and in our work spaces. But in the datacentre and service provider environments optical networking is the predominant deployment especially now that 10G/25G/100G networking is more pervasive.

Most networking engineers including myself do not have a good understanding of optical networking. This often leads to confusion, misconceptions and errors when working fibre optic networking.

Optical fibre cables are just a medium to transport light from one end to another. The material is typically made from fiberglass or plastic and there are many reasons why fibre optics are used in networking. Some of these include;

• Relatively lower cost compared to copper
• Capability to carry signal over very long distances
• Capability to carry extremely high bandwidth signals
• Less interference

Optical fibre carries light signal due to total internal reflection. Fibre optic cables are internally composed of two layers

• A core, surrounded by a different material known as the cladding
• The cladding always has higher index of refraction than core
• When the light tries to pass from the core to the cladding and the angle is correct it is reflected back to the core

Fibre optic cables are typically deployed in a duplex mode, i.e. there is a strand for transmitting and receiving. A fibre optic cable typically carries signals generated by transceiver optics connecting to x86 based devices, routers, switches and other networking devices. Digital signals from these devices is then encoded into analogue pulses by the optics connecting to these devices. The receiving device’s optic transceiver then in turn changes the analogue pulses in to digital signals.

As laid out in Richard’s presentation there are few methods of encoding these digital signals in to analogue pulses;

• Intensity Modulation with Direct Detection (IMDD) which is encoded as Non-Return to Zero (NRZ). NRZ is a fancy way of saying bright for a 1, dim for 0. This modulation is called baud can happen billions of times/sec. The receiving end sees flashes and turns it to 1s and 0s. This technique was used for all optical signals up to 10Gbps. Beyond 25GBaud, this technique gets increasingly hard to scale
• QPSK 100G is basis of most long-haul links today
• 100GBASE-PAM4 (Pulse Amplitude Modulation) QSFP28 optics delivering 80km 100G cheaply are starting to ship today, etc. QSFP stands for Quad Small Form-factor Pluggable, this is a hot-pluggable transceiver used for data communication applications.

There are two basic types of fibre optics, multi-mode and single-mode.

• Multi-Mode Fibre

  • This is relatively lower cost compared to single-mode fibre
  • It used in short reach deployments, most of the intra datacentre connections, i.e. server to switch, or switch to router connections leverage short reach optics using multi-mode fibre
  • Multi-Mode allows multiple modes of light to propagate through it
  • There are couple types of MMF
    • OM1/OM2
      • Orange coloured fibre cables are in this range
      • OM1 has 62.5 micron (um) core, OM2 has a 50um core.
      • Failure rate goes up with anything beyond 10G
    • OM3/OM4
      • Aqua coloured fibre cables
      • Supports 10G signals at much longer distances (300-550m)
      • 40G/100G signals require this

• Single-Mode Fibre

  • It used for high bandwidth and long distances
  • It has smaller core size compared to multi-mode (8-10 microns)
  • Due to the smaller core size there are no inherent distant limitations caused by modal distortions
  • SMF has two standards
    • OS1, usually referred to as tight buffered and used indoors
    • OS2, usually referred to as loose and used outdoors for buried use

When it comes to optical networking there are some terms and concepts that we need to familiar with. Richard does a nice job in laying these out.

• Optical power
  • This is this the brightness of light
  • As light travels through Fibre some energy is lost
  • Richard also points out when measuring light at the end of a fibre optic cable it is important that we know the difference between dB and dBM.
• Dispersion
  • This typically means that as light propagates through the fibre cable there will be an inherent signal degradation. There are two main types of dispersion
    • Chromatic dispersion
      • In chromatic dispersion when one part of an optical signal travels faster than the other part, the signal will eventually smear out over long distances
    • Polarisation mode dispersion
      • This is caused by imperfection in shape of the Fibre, e.g. bent fibre,
• Fibre Optic Transmission bands
  • There are several frequency bands also referred to as windows
    • 850nm – The first window
      • Used on MMF
      • Highest attenuation, only used for short reach applications
    • 1310nm – The second window (O-band)
      • Used on SMF
      • Has high attenuation and primarily used for medium reach application (10km)
    • 1525-1565nm – Third window (C-band)
      • Used on SMF and for almost all long-reach and DWDM applications
      • This has the lowest rate of attenuation over SMF
    • 1570-1610nm – Fourth Window (L-band)
      • Typically used in under sea cables

Today optical networking powers the Internet. This basically allows communication between Internet infrastructure deployed in different parts of town, city, country or continents. Vast array of land buried and under sea fibre cables connects countries and continents. These fibre optic cables are even carrying the data served by this WordPress instance right to your web browser. Unlike our dedicated copper cables connected to our computers, these fibre optic cables carry yours and many other user and machine generated traffic. In order to carry these digital signals a method called Wavelength Division Multiplexing or WDM in short is used. This is basically combining different light sources at different wavelengths so that they can be carried over a shared fibre optic cable.

• There are two types of WDM systems
  • Dense WDM (DWDM)
    • DWDM has more tightly packed WDM
    • It operates within C-band and the following channels are common
      • 200GHz – 1.6nm spacing, 20 channels
      • 100GHz – 0.8nm spacing, 40 channels
      • 50GHz – 0.4nm spacing, 80 channels
      • 25GHz – 0.2nm spacing, 160 channels
    • 50GHz is the commonly deployed for commercial and long-haul 100G systems
  • Coarse WDM (CDWDM
    • It has 8 channels with 20nm space
    • With Lower Water Peak Fibre, another 10 channels are possible
  • Both systems do the same thing, the difference is the channel spacing and sometimes the range of the optical spectrum they cover. DWDM systems are however more popular due to it’s capabilities.

A typical WDM system has the following main components

• Mux/Demux
  • Short for multiplexer/demultipler
  • Simple device which combines (Mux) or splits (Demux) multiple colours of light into a single Fibre
  • Muxes are passive device, requiring no power
• Optical Add/Drop Multiplexer (OADM)
  • This component selectively adds and drops certain WDM channels while passing other channels through without disruption
  • Where a Mux is used at the end-point of a WDM network an OADM is used at a mid-point, often in a ring
  • A well constructed OADM ring can reach any other node in the ring, potentially reusing the same wavelenght multiple times across different portions of the ring

• Reconfigurable OADM (ROADM)

  • This is a software tunable OADM that allows you to control which channels are dropped and which are passed through, thus increasing flexibility

• The Evolution of the ROADM

  • Basic ROADM
    • Reconfigurable but Add/Drop still goes to a standard fixed Mux
    • Specific frequencies must be connected specific ports
    • The network must be recabled in order change or move frequencies
  • Colourless ROADM
    • Eliminates the need to map specific frequencies to specific ports
    • But still limited to Muxing in one direction
  • Colourless Directionless Contentionless (CDC) ROADM
    • Any channel can be add/dropped on any port
    • Any channel can be sent to any direction
    • The same channel can be reused on different direction without causing internal contention in the ROADM

Vendors such as Infinera, Adva, Ciena, Huawei manufacture DWDM systems that are used to transport different multiplexed wavelengths across long distances.

When engineering an optical network it is important to understand that fibre is susceptible to insertion loss and that even the best connectors and splices are not perfect. It is also important to know that over period of time fibre optic cables and transceivers start to degrade and the appropriate budgets to should be put in place to manage the life-cycle of the fibre optic infrastructure.

The current technology progression within optical networking is seeing developments around coherent optical technologies. Basically these are group of advancements in optical technology which are combined together to significantly increased optical performance. Coherent optical technologies consist of;

• Polarisation multiplexing
• High-order phase modulation techniques
• Using laser as a local reference oscillator on the receive side
• Deployment of advanced Digital Signal Processors which tie of all these together
• True 100G/200G and beyond optical signals, not just Nx10G signals

I highly encourage going through Richard’s Nanog presentation, I feel like I have just scratched the surface on optical networking and there many more things to learn. I will attempt to release another series within optical network that is a bit more focussed in the coming future.