Looking back on the optical Ethernet standards so far and the background of the discussion over "40GBASE-FR" [Net new technology]-INTERNET Watch

"40GBASE-FR" suitable for PCS using 64B/66B Encode with a reach of 2km

So this time we will talk about "40GBASE-FR". This standard was standardized as "IEEE 802.3bg-2011" in 2011. It is included in IEEE 802.3 as Clause 82/89. Clause 82 is a specification for PCS that uses 64B/66B Encode, and it uses "40GBASE-R" and "100GBASE-R", which is basically the same specification as "100GBASE-SR4".

What is different is the signal speed, which is defined in Clause 89. The wavelength is 1530 to 1565 nm, and one lane is used for transmission and reception. The signal speed is 41.25GBd, and through 64B/66B Encode, the data transfer rate is exactly 40Gbps.

As you can see from the FR, the reach is 2m to 2km. Naturally, it must be used in SMF. By the way, the center wavelength is 1550 nm. It seems that the use of 1310nm was also considered in the initial stage of consideration, but this was abandoned at an early stage.

　The original purpose of 40GBASE-FR was to use existing 40G standards to formulate 40G Ethernet standards quickly and at low cost. The existing 40G standard is OTU3/STM-256/OC-768/40G PCS (Packet over SONET), and this physical layer is used as it is for Ethernet. The troublesome thing was that OTU3/STM-256 was specified as ITU G.693 and G.959.1, but G.693 listed the following two wavelengths. We argued over whether to use it or not.

By the way, the advantages and disadvantages of both were summarized as follows.

Advantages of 1310nm:

Disadvantages of 1310nm:

Advantages of 1550nm:

Disadvantages of 1550nm:

For this reason, at the first meeting of the Task Force, there was a proposal to use 1310nm.

However, in the end, 1550nm was selected due to the balance of power consumption, etc., and it seems that standardization was pushed forward as it was. The Task Force held its first meeting in May 2010, finished its last meeting in March 2011, and completed standardization in 2011.

　Compared to other standards, there is a feeling that the work progressed terribly quickly, but was it widely used? But it wasn't. The reason may be that the standard was half-finished.

40Gbps/2km with a single pair of optical fiber was certainly not the only option at the time, but ``40GBASE-SR4'' for short distances and ``100GBASE-LR4'' for long distances had already been standardized. It's over, and if 40G is fine, "40GBASE-SR4/40GBASE-LR4" products are on the market, and there was no merit in switching to 40GBASE-FR now.

Also, the fact that the module was CFP instead of QSFP+ etc. was already a disadvantage. Looking back, there was a standard called "10GBASE-FR" first, and if we could aim for this upgrade, we might have been able to demonstrate a little more presence, but that's why 40GBASE-FR will disappear without being used much. has become

"10GBASE-T, finally spreading?" List of articles

[Road to 10Gbps access line] List of articles