QSFP-DD vs OSFP: The Wave of the Future 400G Transceiver

Over the past years, the discussion of 200G and 400G Ethernet speeds started heating up among data center professionals. The next station for IT community is 400G. And two standards organizations have been working behind the scenes with this benchmark for the past couple years, and in 2016 the long-awaited public launch of efforts to develop QSFP-DD and OSFP optical transceiver modules for 400-Gbps applications had finally arrived. Although 400G Ethernet is not the mainstream now, but we can still not deny the fact that it will finally become the mainstream in the near future. And QSFP-DD vs OSFP optical transceiver will be the top trend for future transceiver.

Introduction to QSFP-DD

QSFP-DD refers to Quad Small Form Factor Pluggable Double Density.  We know that current QSFP28 optical modules support 40 and 100 Gigabit Ethernet applications. They feature four electrical lanes that can operate at 10 or 25 Gbps. While the new QSFP-DD is designed with eight lanes that operate at up to 25 Gbps via NRZ modulation or 50 Gbps via PAM4 modulation, which would support optical transmission of 200 Gbps or 400 Gbps aggregate, thus doubling the density. At the 400G transmission rate, the QSFP-DD form factor could enable up to 14.4 bps aggregate bandwidth in a single switch slot so as to cope with rapid data center traffic growth. Another greatest advantage of this new generation of QSFP is its backward compatibility with the QSFP and QSFP28.

Introduction to QSFP-DD

Introduction to OSFP

OSFP stands for Octal Small Formfactor Pluggable, which is a very new module and interconnect system in development that is targeted to support 400G optical data links inside data centers, campuses and external metro long reach. OSFP’s first iteration is an 8-lane times 50G PAM4 = 400G physical link with a possible future 4 x100G PAM4 = 400G and 8 lanes times 100G PAM4 = 800G variants. This physical packaging system is agnostic relative to the different protocol I/O interfaces that will likely use it. The OSFP module or cable plug is a dual paddleboard direct-attachment type connection. This form factor allows 32 400 Gb/s ports per 1U to enable 12.8 Tb/s per switch slot. And the OSFP to QSFP+ adapters will support backward compatibility between form factors.

Introduction to OSFP

400G QSFP-DD vs OSFP: Who Will Win?

OSFP’s module size is said to be slightly wider and deeper than the QSFP-DD module, thus taking up more PCB surface area. Several OSFP modules on a line card use much more area and only 32 ports per 1U box faceplate are possible versus the 36 ports of QSFP-DD. The larger size OSFP may not have enough power and cooling advantages versus QSFP-DD’s density and its four extra port capability. Different users with different applications may fervently prefer one connection system versus the other relative to their panel density, cost, performance, power and cooling priorities. And according to the market forecast, the QSFP-DD should be ready before the OSFP, which is another advantage in addition to its backward compatibility with QSFP and QSFP28. As it has been mentioned, OSFP will need an adapter to support backward compatibility between form factors. It seems that the future for QSFP-DD is brighter. But now it is still too early to draw the conclusion and the two modules will have slightly different applications .

Summary

Although many data centers are still upgrading their network to a 100G data rate thanks to the cost-effective and high performance, 400G optical transceivers, as the wave of the future transceiver, are already on the way. As for the question: 400G QSFP-DD vs OSFP: Who Will Win? Let’s wait and see!

Posted in Fiber Optic Transceivers | Tagged , | Comments Off on QSFP-DD vs OSFP: The Wave of the Future 400G Transceiver

Besides the Chip, What Else Composes an Ethernet Switch?

“The chip is the heart of a switch.” This is already known to us. But a chip cannot work alone to allow the operation of a switch. Other familiar parts include the noisy fans that often reminds us the switches are working hard, and the power supplies that power up the switch. Then what about the operating systems? What else does an Ethernet switch consist of?

What Does an Ethernet Switch Basically Have?

If taking an Ethernet switch apart as small as possible in mechanical level, there can be hundreds and thousands of components. But it is not the purpose of this post. In a simple way a switch basically consists of six components. See the table below:

Table: switch components in a simple way.

Chassis The metal part that contains other components. Have personal color and assembly options.
Power Supplies 1+1 redundant. They light up the switch.
Fans There should be enough fans to cool down and they are also the evidence that the switch is alive.
Fan control PCBA PCBA (Printed Circuit Board Assembly) where fans are connected.
CPU PCBA The x86 or Power PC or ARM based processors that has it’s RAM, FLASH and PCIe that runs the switch OS.
Switch main board PCBA The board that hosts the main switch silicon, interface cages, CPLDs (Complex Programmable Logic Devices) and PHYs (in case of RJ45 interfaces), this PCB is between 16 to 22 layers.

Here is the top & front view of a Nephos NPS4806 48x10G & 6x40G switch for illustration. You can easily found the six basic components. It is based on MediaTak/Nephos Aries MT3258 silicon, which is protected by the middle fat heat sink. These Application Specific Integrated Circuits (ASICs) are packaged into ball grid array (BGA) and are normally large in size, because they have many interface pins underneath the silicon.

Ethernet Switch

The CPLDs are responsible for system startup, managing the LEDs, fans and temperature sensors. The diagram of the main PCB board shows how different components are connected in the switch. The CPLDs are connected back to the CPU through the UART (Universal Asynchronous Receiver/Transmitter) or I2C (IC to IC Interconnect) bus.

Nephos main PCB board

Looking at the CPU that is based on Intel’s Atom C2538 and the MT3258 chip, there is a direct PCIe connection between them. This connection requires a Driver. You can try using OF-DPA (OpenFlow Data Plane Abstraction) on ONL (Open Network Linux) to drive this connection. Most of today’s switches are based on Broadcom chip, and the Driver is provided by the Broadcom SDK to the NOS (Network Operating System) vendors that they can include in their OS.

There are three CPLDs on this switch. The CPLD on the top controls the fans, temperature and LEDs. This CPLD has an expose I2C interface which is connected to the CPU board. The CPLD2 & CPLD3 are for controlling the SFPs. This switch has no on-board PHY interface, and uses the 10/1G SFP1 for interfaces. The CPLD2 &CPLD3 control the SFPs, TX Fault, TX Disable, RX LOSS and Mode. Each controls 24 SFPs.

What’s the Difference Between Different Switches?

Though there are certain components that are fixed in all switches, you will find that switches from different manufacturers have both many similarities and differences. They may be using the same Broadcom chips of Strata XGS Family or Strata DNX Family, however, their CPU board, CPLDs, and also commands used to read and write the I2C bus are different. That’s why porting the NOS between platforms is not easy and usually requires extensive testing. In short, the CPLD commands and I2C addresses might be different between such platforms.

Here are some pictures of several 10G, 25G, 40G and 100G switches based on Broadcom chips. The main difference that can be visually is the layout and the components used.

10G Switch

Juniper OCX1100-48SX is a 48x10G and 6x40G switch. It follows standard layout. And the left pic just presents some major components and the right pic displays some more details of the PCBAs.

Juniper OCX1100-48SX 48x10G & 6x40G top view

40G Switch

Edge-Core AS6712 with 32x40G ports.

Edge-Core AS6712 32x40G top view

25G Switch

Edgecore Networks AS7300-54X with 48x25G and 6x100G ports.

Edgecore AS7300-54X 48x25G & 6x100G top view

100G Switch

Alpha Networks SNH-60×0-320F switch with 32x100G ports.

Alpha Networks SNH-60x0-320F 32x100G top view

What About the Brother of the Hardware?

Of course an Ethernet switch is not complete with only hardware. The NOS installed also makes a difference. Some switches support only a certain type of NOS while some switches are “open”. These switches support an environment called ONIE (Open Network Installation Environment) and are known as white-box switches. An open networking switch support several compatible NOS over ONIE, such as Cumulus Linux, IP Infusion, Pica8, Open Network Linux, etc. With both the hardware and software ready, the switch can begin its work.

Summary

Now you can say that, well, making an Ethernet switch is just as simple as a BB8 model from Star Wars. You can complete one by yourself with ready-made materials. And if you are a master about switch, you may do better than just combining them together. This post is just for sharing some basic knowledge about switch. More technical knowledge about Ethernet switch can be obtained via tech@fs.com if you’re interested.

Posted in Enterprise Network | Tagged , , , | Comments Off on Besides the Chip, What Else Composes an Ethernet Switch?

SFP-10G-LR vs SFP-10G-SR: Which One to Choose?

Although 40G and 100G optical transceivers are on the very top trend for enterprise and data center for the interconnection since 40G/100G network is racing to the future, 10G SFP+ optical transceiver is still considered as the mainstream form factor of the 2017 market due to its matured technology and lower price. There are various  SFP+ modules that offer 10 Gigabit Ethernet connectivity including SFP-10G-SR, SFP-10G-LR, 10GBASE-ER SFP+, and 10GBASE-LRM SFP+. Here, we mainly deals with two of them, SFP-10G-LR vs SFP-10G-SR.

An Overview of Cisco SFP-10G-LR and SFP-10G-SR

Before talking about Cisco SFP-10G-LR and SFP-10G-SR, let’s have a look at the Cisco 10G SFP+ modules first. Cisco 10G SFP+ modules contain the following features and benefits: it supports 10GBASE Ethernet with the smallest 10G form factor; with hot-swappable input/output device, it can be plugged into an Ethernet SFP+ port of a Cisco switch (no need to power down when installing or replacing); its digital optical monitoring capability ensures strong diagnostic capability; and it also provides much more flexibility and convenience for interface choices.

SFP-10G-LR vs SFP-10G-SR

Cisco SFP-10G-SR transceiver is compliant with 10GBASE-SR standard. It is hot-swappable input/output device which allows a 10 Gigabit Ethernet port to link 300m in a fiber optic network. Because it is hot-swappable and MSA compliant, the Cisco SFP-10G-SR transceiver can be plugged directly into any Cisco SFP+ based transceiver port, without the need to power down the host network system. This capability makes moves, add-ons and exchanges quick and painless.

While Cisco 10GBASE-LR module can support a link length of 10 km on standard single-mode fiber. Compliant with 10GBase-LR standard, SFP-10G-LR can support up to 10km over single-mode fiber and uses 1310nm lasers. There is no minimum distance for LR, either, therefore it is suitable for short connections over single mode fiber, too. The following table shows a datasheet of Cisco SFP-10G-SR and SFP-10G-LR:

Product Name

SFP-10G-SR

SFP-10G-LR

Data Rate

10.3 Gbps

10.3 Gbps

Wavelength

850nm

1310nm

Distance/Power Budget

300m

10km

Output Power

-7.3~-1.2dBm

-8.2~0.5

Receiver Sensitivity

-11.1dBm

-14.4dBm

Power Supply Voltage

3.3V

3.3V

Connector

Dual LC

Dual LC

Fiber Type

MMF

SMF

Operating Temperature

Dual LC

Dual LC

Application

10 Gigabit Ethernet

10 Gigabit Ethernet

Form Factor

SFP+

SFP+

Digital Optical Mon (DOM)

Yes

Yes

Difference between SFP-10G-LR and SFP-10G-SR

This difference between SFP-10G-LR and SFP-10G-SR is pretty simple. SR stands for Short Reach, and LR stands for Long Reach. SR transceivers are almost always multitude, and optimized for high speeds over relatively short distances. However, the much higher-powered OM3 and OM4 formats can push that into the hundreds of meters for a single cable. While LR transceivers are designed for long-range communications, such as wiring buildings together on a large campus or even setting up a Metro Area Network (MAN). They can be either multitude or single-mode, and are almost always intended for 100m+ applications. Higher end cables and transceivers can support transmissions of several kilometers

FS Cisco Compatible SFP-10G-LR and SFP-10G-SR Products

To buy Cisco original brand transceiver may cost you a lot of money, and the question is that do you really need the Cisco original brand modules for your Cisco switch. To be frankly, not every SFP+ modules can work well on the Cisco switch SFP+ ports. You must make sure of the compatibility before plugging the SFP+ transceivers on it. At FS, we provide high-quality Cisco compatible SFP-10G-SR and SFP-10G-LR modules at a low price. And we can assure you that every piece of transceiver is individually tested on a full range of Cisco equipment and passed the monitoring of our intelligent quality control system. Part of the products are listed below.

Part Number

Description

Price

11552

Cisco SFP-10G-SR Compatible 10GBASE-SR SFP+ 850nm 300m DOM Transceiver

US$ 16

11555

Cisco SFP-10G-LR Compatible 10GBASE-LR SFP+ 1310nm 10km DOM Transceiver

US$ 34

58773

Generic Compatible 10GBASE-SR SFP+ 850nm 300m DOM IND Transceiver

US$ 16

11591

Generic Compatible 10GBASE-LR SFP+ 1310nm 10km DOM Transceiver

US$ 34

Summary

After the comparison, have you got the answer for the question, SFP-10G-LR vs SFP-10G-SR: Which One to Choose? Actually it totally depends on your actual needs. As a reliable and qualified fiber optics supplier, FS is your ideal choice for Cisco compatible SFP-10G-LR and SFP-10G-SR transceivers. For more details, please visit www.fs.com

Posted in Fiber Optic Transceivers, SFP+ | Tagged , , | Comments Off on SFP-10G-LR vs SFP-10G-SR: Which One to Choose?

How to Understand Link Budget and Link Loss in Fiber Optic Network?

If you’re a network engineer once involved in a cable plant installation project, you must have heard the term Link Budget. People in this area know how important it is to a fiber optic network cabling. During the design stage of the cabling, link budget is adopted to predict the amount of light required to ensure an uninterrupted communications link. And another closely related term is Link Loss Budget. Together they contribute to the proper operation of a fiber run.

Link Budget, Link Loss and the Margin

Link budget, or power budget, refers to the amount of loss that a data link (transmitter to receiver) can tolerate in order to operate properly. Sometimes it has both a maximum value and a minimum value, so that the input power at the receiver end is within its operating range.

Link loss budget is the amount of loss that a cable plant should have. It is calculated by adding the losses of all the components used in the cable plant to get the estimated end-to-end loss. Obviously, the link budget and link loss budget are related. A data link will only operate properly when the link loss is within the link budget of the link.

The difference value between the power budget and the link loss budget is known as link budget buffer. The buffer value should not be too small, because the margin for error is 3 dB in a fiber link. If the in-between components are fixed, then you can save more margin by changing the transmitter or receiver on two ends; if the two end devices are fixed, you can save yourself more margin by changing the fiber optic jumpers and other passive components.

How to Calculate the Link Budget and Link Loss Before a Cable Plant?

Generally, four main parameters are used to calculate the optical transmission link budget buffer. They are minimum optical transmitter power, maximum connector insertion loss, optical fiber cable transmission loss and maximum optical receiver sensitivity.

Transmitter power and receiver sensitivity are absolute values (e.g. mWatt or dBm, 10*log(mW) = dBm), but connector insertion loss and optical fiber cable transmission loss are relative values (e.g. % loss). The connector insertion losses comprise the connections of fiber optical jumpers, transceivers, patch panels, etc. In order to help understand how to calculate the link budget, here is an example of a typical 2 kilometer multimode link with 5 connections (2 connectors at each end and 3 connections at patch panels in the link) and one splice in the middle. The maximum fiber loss of multimode fiber is 3.5 dB/km and the maximum acceptable connector insertion loss is 0.75 dB. Typical splice loss for multimode fiber is 0.3 dB. Therefore, the total maximum link loss is 11.05 dB.

the link budget and link loss illustration

Figure: the link budget and link loss illustration of a typical 2 kilometer multimode fiber link.

Table 1: calculation of link budget and link loss.

Absolute Values Relative Values
Minimum transmitter power (Tx) -5 dBm Optical fiber cable transmission loss 7 dB
Maximum receiver sensitivity (Rx) -21 dBm Maximum connector insertion loss 3.75 dB
Typical multimode fiber splice loss 0.3 dB
Maximum transceiver link budget 16 dB Maximum link loss along the fiber run 11.05 dB
Network segment link budget buffer = 4.95 dB
Will the Optical Fiber Cable Type Matter to Link Loss?

Yes. Different fiber cable types have different fiber loss at their working wavelengths. According to TIA-568-C.0- 2 standards, their maximum values are as below:

Table 2: maximum loss values of multimode fiber and single-mode fiber.

Fiber Type Multimode Single-mode
Wavelength (nm) 850 1310 1550
Fiber Loss/km (dB) 3.5 0.4 0.3
Insertion Loss (dB) 0.75 0.75 0.75
How to Add More Connections in the link?

If you want to add more connections in the link when the two absolute values are known, the simplest way is to choose fiber optic jumpers with low insertion loss as much as possible. Because the connector insertion loss contributes a lot to the total link loss in a fiber link. You can also use more precise splice machines but it is not as easy as using lower loss fiber optic jumpers.

How Much Light Is Left at the End of the Line?

The table below shows the amount of loss and the percentage of light remained. 20 dB of loss equals a loss of optical power in a system of 99%. And 30 dB of loss equals 99.9% of loss. 30 dB is typically the most loss a communications system can have since 10-10 error count cannot be factored with less than 0.1% of light.

Table 3: the amount of loss and the percentage of light remained.

link loss and light remained

Summary

Link budget and link loss budget are both vital analysis measures in fiber optic network design. Link budget is mainly used before the installation, whereas link loss budget is used before and after the installation. After the cable plant is installed, the calculated loss values are compared with the test results to ensure the link can operate properly.

Posted in Enterprise Network | Tagged , , , | Comments Off on How to Understand Link Budget and Link Loss in Fiber Optic Network?

Why Use Tunable DWDM SFP+ Transceivers?

The tunable DWDM SFP+ is one kind of DWDM SFP+ transceivers. They both can be used in the DWDM system. In the market, tunable DWDM SFP+ transceivers are often between two and four times more expensive than DWDM SFP+ transceivers. Thus, many may think DWDM SFP+ transceivers are enough in the DWDM system and wonder why tunable DWDM SFP+ transceivers are also needed. This post will introduce what is tunable DWDM SFP+ transceiver and explain why they need to be used in DWDM systems in details.

Tunable DWDM SFP+ Transceivers

What’s Tunable DWDM SFP+ Transceiver?

Tunable SFP+ transceivers are a new technology that is in development for a few more years due to the limiting power specifications of the SFP+. They are only available in DWDM form because the CWDM grid is too wide. So a tunable SFP+ transceiver is also called tunable DWDM SFP+ transceiver.

The tunable DWDM SFP+ transceiver is equipped with an integrated full C-Band 50GHz tunable transmitter and a high performance PIN receiver to meet the ITU-T (50GHz DWDM ITU-T Full C-band) requirements. It shares the same hot–pluggable SFP+ footprint as DWDM SFP+ transceiver. The major difference between them is that DWDM SFP+ has a fixed wavelength or lambda while the tunable DWDM SFP+ can adjust its wavelength on site to the required lambda. Tunable DWDM SFP+ transceivers enable us to change wavelengths unlimited within the C-band DWDM ITU Grid and can be applied in various types of equipment such as switches, routers and servers.

tunable-transceiver

Why Tunable DWDM SFP+ Transceivers Are Used in DWDM Systems?

In traditional DWDM systems, fixed-wavelength DWDM SFP+ transceivers are commonly used as light sources in optical communication field. However, as the continuous development, application and promotion of optical communication systems, the disadvantages of DWDM SFP+ transceivers have been gradually revealed. The followings are why tunable DWDM SFP+ transceivers are also needed in DWDM systems:

On the one hand, it is essential to prepare backup DWDM SFP+ transceivers for each DWDM wavelength to avoid unnecessary breakdown. In traditional DWDM systems, a small number of extra DWDM SFP+ transceivers are enough. However, with the development of technology, the number of wavelengths in DWDM 50GHz now has reached the hundreds. This means people have to provide up to hundreds of backup DWDM SFP+ transceivers, which will greatly increase the operating cost. Tunable DWDM SFP+ transceivers provide equipment manufacturers and operators with great flexibility, achieving the optimization for the overall network performance and greatly reduce the demand of existing operators for DWDM SFP+ transceiver inventory.

On the other hand, in DWDM systems, it may be required to use a large number of DWDM SFP+ transceivers with different wavelengths to support the dynamic wavelength assignment in optical network and improve network flexibility. But the usage rate of each transceiver is very low, resulting in a waste of resources. The arrival of tunable DWDM SFP+ transceivers has effectively solved this problem. With tunable DWDM SFP+ transceivers, different DWDM wavelengths can be configured and output in the same light source, and these wavelength values and intervals all meet the requirements of ITU-T (50GHz DWDM ITU-T Full C-Band).

Conclusion

Featuring for flexibly selecting working wavelength, tunable DWDM SFP+ transceivers have very large practical value in optical fiber communication wave division multiplexing system, optical add-drop multiplexer and optical cross-connection, optical switching equipment, light source parts and other applications.

Posted in CWDM/DWDM Transceivers | Tagged , | Comments Off on Why Use Tunable DWDM SFP+ Transceivers?