Data Center

Checkpoint No. 3 for Fiber Infrastructure Deployment: Optical Fiber Standards

Qing Xu
To support the expanding cloud ecosystem, optical active component vendors have designed and commercialized new transceiver types under multi-source agreements (MSAs) for different data center types; standards bodies are incorporating these new variants into new standards development.

For example, IEEE 802.3 taskforces are working on 50 Gbps- and 100 Gbps-per-lane technologies for next-generation Ethernet speeds from 50 Gbps to 400 Gbps. Moving from 10 Gbps to 25 Gbps, and then to 50 Gbps and 100 Gbps per lane, creates new challenges in semiconductor integrated circuit design and manufacturing processes, as well as in high-speed data transmission.

 

As you get ready for new fiber infrastructure deployment to accommodate these upcoming changes, there are four essential checkpoints that we think you should keep in mind:

 

  1. Determine the active equipment I/O interface based on application types
  2. Choose optical link media based on reach and speed
  3. Verify optical fiber standards developed by standards bodies
  4. Validate optical link budget based on link distance and number of connection points

 

Optical fiber standards are developed by standards bodies, such as TIA and ISO/IEC, to support various applications with rigorous specifications and guidance, and to ensure a healthy industry ecosystem.

 

Multimode Optical Fiber Standards

The original multimode (MMF) optical fiber standard, TIA-492AAAA (OM1, 62.5/125 µm), was released in 1989 to support Fast Ethernet 100BASE-FX and 1000BASE-SX Ethernet, with a high NA of 0.275, and better capture light from 1300 nm LED sources.

 

Then, the TIA-492AAAB optical fiber standard for OM2 (50/125 µm) with improved modal bandwidth and reduced NA of 0.2 was released in 1998 to support higher data transmission, such as 1 Gbps VCSEL with longer reach.

 

Although OM1 and OM2 have been widely deployed, they are no longer suitable for new Ethernet infrastructure deployment. The overfilled launch condition was a characterization of LED-based systems in which the light was launched over the entire core of the fiber and then measured at 850 nm and 1300 nm.

 

To meet growing bandwidth requirements, laser-optimized multimode fiber (LOMMF) standards OM3 and OM4 were developed in 2002 and 2009, with effective modal bandwidth (EMB) of 2000 MHz∙km and 4700 MHz∙km to support 10G, 40 Gbps and 100 Gbps Ethernet applications, as well as InfiniBand and Fibre Channel protocols.

 

While the bandwidth requirements are specified at 850 nm and 1300 nm, 850 nm VCSEL-based transceivers are dominant in the market thanks to low cost, high yield and improved EMB of OM3 and OM4 fiber; however, as a multimode laser source, the VCSEL spectral width is also a limiting factor for fiber reach.

 

In June 2016, a new standard for OM5, which is also referred as wideband MMF (WBMMF), was approved and published to support the short-wavelength multiplexing at 850 nm to 950 nm to increase data-transmission rates by a factor of four in one single multimode fiber.

 

fiber-checkpoint-3-Table-1 

Click the image for larger version

Singlemode Optical Fiber Standards

OS1 singlemode fiber (SMF) is a legacy two wavelength-window product. Recently, OS1a was adopted by the ISO/IEC as the indoor tight buffered fiber standard. Loose-tube OS2 is currently the most popular SMF cable type, specifying low water peak and low loss with three operational wavelength windows.

 

fiber-checkpoint-3-Table-2 

Click the image for larger version

 

Hyperscale data centers, such as Amazon, Google, Microsoft and Facebook, have already scaled beyond VCSEL-based transceiver reach limits. To suit the needs of the mega data center footprints and interconnect topology, singlemode transceiver variants, such as 100 Gbps LR4 (10KM), CWDM4 (2KM), CLR4 (2KM) and PSM4 (500M) modules, paired with SMF cabling, can achieve the highest cost efficiency and economy of scale.

 

In addition, cloud data center operators have high negotiating power and initial volume to customize optical transceiver specifications and build solid “engineered links,” which are not necessarily compliant with industry standards.

 

Meanwhile, for the majority of the data center market (Tier 2/3 web portals, enterprises, public organizations and multi-tenant data centers), the cost of standard transceivers still dominates the cost of the link.

 

Typically, the cost of a multimode transceiver is 1.5 to 3 times lower than the cost of singlemode transceivers (and consumes 40% to 60% less power). For that reason, multimode optics are still a favorable choice for short-reach applications due to their cost and power advantages and broad product availability.

 

In the new optical fiber standard, ANSI/TIA-568.3-D, some important MMF and SMF cable specification changes have been made:

  1. Denotes legacy OM1, OM2 MMF and OS1 SMF as “not-recommended”
  2. Adds specification for wideband MMF TIA-492AAAE, which was named OM5 by the ISO/IEC standard body; lowers maximum allowable laser-optimized OM3, OM4 and OM5 MMF attenuation at 850 nm from 3.5 dB/km to 3.0 dB/km;
  3. Lowers maximum allowable loose-tube OS2 SMF attenuation from 0.5 dB/km to 0.4 dB/km
  4. Only the “low water peak” SMF with three operational wavelength windows (1310 nm, 1383 nm and 1550 nm) is recommended for new installations; the tight-buffered SMF successor of the OS1 has been named OS1a by the ISO/IEC standard body
  5. Lowers maximum return loss of singlemode connections and splices from 26 dB to 35 dB

Don’t miss the last blog in this series, where we will elaborate on MMF and SMF cable solutions, and provide practical guidelines for understanding the fiber link budget for new fiber infrastructure.

 

Belden can help you with fiber infrastructure deployment, whether it’s being installed in a new or existing building. We’ll help you consider all of these checkpoints to design a data center solution that provides the speed and longevity you need.