Welcome to gbic-shop.de/blog


Ethernet industry and networking are working in another direction, and it is 25 Gigabit Ethernet with the increasing demand in a cloud database for bandwidth. If we compare the advanced ways of 10G Ethernet-25G Ethernet-100G Ethernet and 10G Ethernet-40G Ethernet-100G Ethernet, we will see that 25G Ethernet seems more refined and popular by customers. Why should we select 25GbE? How can we interpret the strong points of 25GbE? This writing will give an outlook of 25GbE from all sides.


Previously, the network engineers were surprised at the concept of a 10GbE connection. But, cloud computing and virtualization generated modern challenges of networking demanding additional bandwidth. Usually, the most significant TOR switches in data centers develop 10GbE very quickly. Moreover, to compete with the requirements, the IEEE approved a 100GbE and 40GbE standard, however for cloud providers, 10GbE is not low-cost or effective in TOR switching, and the implementation of 100GbE is comparatively expensive and complex. 25GbE standard was created in opposition to such a situation by IEEE 802.3, utilized for the connectivity of Ethernet that will provide benefit to enterprise and cloud database conditions. The foundation of 25G Ethernet is on IEEE 100G Ethernet, and it utilizes one lane 25Gbps connections of Ethernet via 4x25 Gigabit per second lanes.


Two major form factors QSFP28 and SFP28, are supported by the physical interface standard of 25G Ethernet. Generally used optical transceivers are 25G Ethernet SFP28.

The 25G Ethernet PMDS identify low price TWINAXIAL copper wires for the function of 25Gbps demanding only two pairs of twin axial wires. Servers can be connected to TOR switches by the connections founded on copper TWINAXIAL wires and work for inter-rack links between routers and switches. Fan-out cables can connect to speeds of 10/25/40/50 Gigabit per second and can right now be achieved on copper cables, SMF and MMF, keeping up with the range to the particular implementation requirements. Generally utilized cables are 25G Ethernet AOC and 25G Ethernet DAC.


When for most of the existing implementations, 10GbE is good, it does not effectively deliver the required bandwidth. Nevertheless, it needs other devices, remarkably increasing costs. Moreover, 40G Ethernet is power efficient and beneficial for cloud providers in TOR switching. So, 25G Ethernet was created for progress and to get out of the difficulty.


SERIALIZER/DESERIALIZER (SERDES) is a combined transceiver or circuit utilized in high-level speed transmissions for transforming serial data to aligned interfaces and likewise. The transponder part is a series to a parallel transformer, and the adopter part is equal to the series converter. Presently, 25Gbps is the rate of SERIALIZER/DESERIALIZER. We can say that at 25Gbps, we require only a single SERDES lane to link one ending point of the 25G Ethernet card to the further ending point of the 25G Ethernet card. Contrarily, four 10G Ethernet SERIALIZER/DESERIALIZER lanes are required by 40G Ethernet to attain links. Consequently, we need four pairs of optical fibers for connections between two network cards of 40G Ethernet. Moreover, 25GbE presents a simple advanced way to 50 Gigabit and 100 Gigabit network.


Popular Intel Xeon CPU presents 40 lanes of PCI Express 3.0 with one lane bandwidth of around 8Gbps. We use the lanes called PCI Express for connections between NIC and CPU. Moreover, we also use these lanes for connections between GPU cards, RAID cards, and all remaining peripheral cards. That's why it is essential to extend the usage of confined PCI Express lanes by NIC. One 40G Ethernet requires not less than a single PCI Express 3.0x8 lane. Therefore, if two ports of 40G Ethernet can work with perfect speeds simultaneously, the usage of actual lane bandwidth is 40 Gigabit* 2/(8 Gigabit*16)=62.5%. Contrarily, 25G Ethernet NIC card requires single PCI Express 3.0x8 lanes, and the usage effectiveness is 25 Gigabit* 2/(8 Gigabit*8)=78%. 25G Ethernet is considerably more flexible and effective than 40G Ethernet regarding PCI Express lanes.


40G Ethernet switches and cards use QSFP+ modules with comparatively expensive MPO/MTP wires not adaptable to LC fiber optics of 10G Ethernet. In case of enhancement to 40G Ethernet founded on 10G Ethernet, many optical fiber wires will be discarded and renewed, which will be very costly. Contrarily, 25G Ethernet switches and cards use SFP28 optical transceivers and are adaptable to LC fiber optics of 10G Ethernet because of a one-lane link. We can evade the renewal of wires in case of enhancement from 10G Ethernet to 25G Ethernet, which proves to be economical and time-saving.


At first, 25G Ethernet has the most significant switch INPUT/OUTPUT presentation and fabric capacity. Network bandwidth presentation of 10G Ethernet can be enjoyed 2.5 times by the cloud companies and web-scale. Transferred across one lane, 25G Ethernet also puts forward network expandability and a higher density of switch ports. In the second place, 25G Ethernet can decrease operating and capital expenses by considerably reducing the desired number of cables and switches accompanied by facility expenses related to power, cooling, and space compared to the technology of 40G Ethernet. Lastly, 25G Ethernet making use of one lane 25 Gigabit per second Ethernet connection protocol leverages the current standard of IEEE 100G Ethernet, which we deploy as four lanes of 25Gbps working on four copper or fiber pairs.


In the past, 25GbE has gained a lot of popularity, and the products of 25G Ethernet have experienced considerable progress and got an expanding market share. In 2020, 25G Ethernet was supposed to find a broader market and continue flourishing in upcoming years. 25G Ethernet is reliable in networks of data centers having high-speed as an adapter of 25G Ethernet can also work at 10G Ethernet speeds. Moreover, 25G Ethernet switches provide a more appropriate way to move to 100 Gigabit or 400 Gigabit networks by getting around the 40G Ethernet upgrade. At the same time, we cannot miscalculate the requirement for industry concord building. Currently, we primarily utilize 25G Ethernet for the implementation switch-to-server. 25GbE can become advanced in case of switch-to-switch implementations are promoted on a large scale. In short, the movement towards 25G Ethernet is progressive.


Regardless of market research, 25GbE inevitably looks to be the desired option because it is cost-effective, offers high bandwidth, and needs low power consumption. Considering the practical advantages of 25GbE, we can predict to proceed with 25GbE in the upcoming days.

A Comparison between G.652 Single Mode Fiber and G.655 Fiber:

Correct measurement and examination in fiber optic wire installation are significant in establishing overall network performance and integrity. A crucial signal loss may cause unstable transmission. How to know the worth of losses upon the fiber connection? This article will guide you to figure out the failures and examine the fiber connection performance.

Types of Optical Fiber Losses:

How can we define fiber loss? Various factors cause light loss, like connector loss, bending, intrinsic element absorption, etc.
We can also call fiber loss attenuation loss or optical fiber attenuation, which counts light loss between output and input.

Fiber optic losses contain dispersion loss, scattering loss, and absorption loss. Structural defects generate them, and Extrinsic Fiber Optic Losses comprise bending loss, connector loss, and splicing loss. We can categorize fiber optic losses into extrinsic and intrinsic ones based on whether the cause of failure is operating conditions or inherent fiber optic characteristics.

Standards for Optical Fiber Loss:

TIA (Telecoms Industry Association)/EIA (Electronics Industry Alliance) develops the standards TIA/EIA, which determine performance and communication requirements for optical fiber connectors, cables, etc. We widely accept and use them in the fiber optic industry. The highest attenuation is the coefficient regarding optical fiber cable, and we can express it in decibel/kilometer units. It is among the significant parameters for the measurement of fiber loss. According to EIA/TIA 568, we have shown the highest attenuation for various types of optical fiber cables in the graph below:

In What Way Can we Calculate Optical Fiber Losses?

To observe if the connection runs appropriately, we must perform the given calculation.

Calculating Fiber Optic Losses:

It is generally the situation to calculate the highest signal loss beyond a given optical fiber connection during the fiber optic cable installation. Firstly, you should know the fiber optic loss formula:

The Total Connection Loss = Splice Loss + Connector Loss + Cable Attenuation

Splice Loss (decibel) = Allowance of Splice Loss (dB) x Amount of Splices

Cable Attenuation (decibel) = Highest Attenuation Coefficient (decibel/kilometer) of Cable x Length (kilometer)

Connector Loss (decibel) = Allowance of Connector Loss (dB) x Amount of Connector Pairs

As the formulas indicate, the total loss is the highest amount of the poor variables in a fiber optic segment. We should note that this method's total failure calculation is only an assessment that presumes the available value of intrinsic losses. Therefore, there is a possibility of higher or less actual loss depending on different factors.

Let's have a practical example to signify the steps of calculation. There is the installation of an SMF cable between two places with a range of 10 kilometers 1310nm fiber optic wavelength. The cable possesses one splice and two pairs of a connector.

  • Calculate attenuation loss of fiber cable. The graph above shows that 1310 nanometer's light attenuation SM outer fiber optic cable is 0.5dB/km. So, the total attenuation of cable is 0.5dB/km x 10 kilometer = 5 decibels.
  • Calculate the complete connector loss. The actual fiber connector loss within practical calculation stands for the value within the optical fiber cable specifications which the suppliers provide.
  • Therefore, we have the total connector failure as 0.75 decibels x 2 = 1.5 decibels. Utilize the TIA/EIA highest loss of one pair as 0.75.
  • Calculate the total splice failure. Utilize the TIA/EIA highest loss like 0.3 for one splice. So, the entire splice loss as 0.3 decibel x 1 = 0.3 decibel.
  • Calculate the more component loss if we have any more components like attenuation.
  • Add the splice loss, connector loss, cable loss, and total connection loss. The entire loss for this connection is 5 decibels + 1.5 decibel + 0.3 decibel = 6.8 decibels.

Note that we have just assumed the estimates. The most accurate and easiest method is to utilize an actual connection's OTDR trace.

Calculation of Power Budget:

How does the value of this connection loss matter for the entire transmission? We will mention here another expression, 'power budget' We use it to compare with the total calculated loss to confirm the cable plant installation is proper. The connection will function typically only if the connection loss is in the loss budget. We calculate the power budget (PB) because of the difference between the output's transmitter within the fiber (PT) and the receiver's sensitivity (PR). We have PB = PT - PR as the calculating formula. Suppose if the output power of an average transmitter is -15dBm, the receiver's sensitivity is -28dBm. We will have power budget as -15 decibel – (-28 decibel) = 13 decibels.

Calculation of Power Margin:

After calculating the power budget and link loss, there is the possibility of calculating the power margin. We call it safety margin. It signifies the power amount attainable after subtracting connection loss from a power budget. We have the formula PM = PB - LL.

Take the case of 10 kilometers SMF as an instance, the connection loss is 6.8dB, and we have a power budget of 13dB. So, 13 decibel – 6.8 decibel = 6.2 decibel is the safety budget. A value higher than zero shows that the connection has enough power for communication.

A Comparison between G.652 Single Mode Fiber and G.655 Fiber:

ITU-T G.65x program is an SMF standard category. We can further divide it into G.652 fiber, G.653 fiber, G.654 fiber, G.655 fiber, G.655 fiber, G.655 fiber, G.656 fiber, and G.657 fiber. Among these, G.655 and G.652 are the options we commonly use. What differences between G.652 and G.655?

Basics and Dissimilarities of G.652 Fiber and G.655 Fiber:

The standardization of the first version of G.652 took place in 1984, and there are four subcategories of this standard: G.652.A fiber, G.652.B fiber, G.652.C fiber, and G.652.D fiber. Among them, G.652.D and G.652.C fibers possess excellent performance compared with G.652.B and G.652.A.All these variants contain a similar G.652 primary size of 8-10µm.

We have given a specification chart regarding four variants of G.652. A to possess not a single dispersion wavelength almost 1310 nanometers-optimal for functioning within 1310 nanometers band. For the divisions of G.652, we have designed G.652.B and G.652. Because of their water peak type, the two are not appropriate for WDM implementations. The more improved variants G.652.D and G.652.C fibers have terminated water peak because of full-spectrum functioning, enabling us to apply them within the wavelength area between 1310 nanometers and 1625 nanometers to help CWDM transmission.

E band refers to the wavelength reach from 1360 nanometers to 1460 nanometers.

S-band stands for the wavelength reaching from 1460 nanometers to 1530 nanometers.

C band refers to the wavelength reach from 1530 nanometers to 1565 nanometers.

L band stands for the wavelength reach from 1565 nanometers to 1625 nanometers.

We can divide G.655 standard into five variants, including G.655.A fiber, G.655B fiber, G.655.C fiber, G.655.D fiber, and G.655.E fiber. We recognize G.655 SMF as NZDSF (Non-Zero Dispersion Shifted Fiber) because of dispersion for a wavelength of 1550 nanometers. We have two kinds of NZDSF: (+D) NZDSF and (D) NZDSF, which represent a positive and negative slope against wavelength.

G.655 fiber possesses a minor, controlled number concerning chromatic dispersion within C-band (1530-1565 nanometers), where amplifiers operate, and it has a significant core size compared with G.652 fiber. G.655, like a developed dispersion-shifted optical fiber, can subdue four-wave merging and more nonlinear effects. That's why G.655 SMF that supports extensive ranges with excessive capacity can encounter the demands of DWDM transmission.