Need Help?

Chat live with us

Live Chat

Want to call?

+1 (888) 468 7419

United States

Country/Region

United States

Language/Currency

English / $ USD

  • English / $ USD
  • Case Study

    See how our customers get inspired by FS solution.

  • Video

    A quick and joyful way to learn about us and our products.

  • Download

    Get software/hardware supporting files for your items.

  • Blog

    Blog articles for data center, enterprise and OTN network.

Cart
Your Shopping Cart Is Empty.

Know-How

fiber optics1
Data Centre

10G - 25G - 100G Network Upgrade: An Inevitable Roadmap for Future Data Centers

10G - 25G - 100G Network Upgrade: An Inevitable Roadmap for Future Data Centers

FS Official 2019-02-12

Driven by the bandwidth requirements of private and public cloud data centers and communication service providers, 25G and 100G experienced significant uptake, resulting in 200G/400G shipments beginning in 2019. Until now, the majority of server vendors have already offered 25 Gigabit Ethernet NICs as the standard I/O option in their servers, driving the latest Ethernet speed transitions from 10G to 25G, 100G and beyond.

Although 1G, 10G and 40G still represent a significant share of the enterprise market’s Ethernet ports, a transition to 25G and 100G has already happened more quickly than ever before. Because the undeniable growth in demand for bandwidth is driving data center networks for better scalability, instant bandwidth provisioning and application agility. This post will highlight the evolutionary path of 25G and why it arrived, then give a comparison over 10G - 40G vs 25G - 100G migration paths. It also clarifies the compatibility issues between 25G and 10G Ethernet standard as well as providing 25G - 50G - 100G - 200G - 400G market highlights for future data centers.

Evolutionary Path of 25G and Why 25G Arrived?

Data centers are expanding at an unprecedented pace, driving the need for higher bandwidth between the server and switches. To cater to this trend, networking and the Ethernet industry are moving from 10G to 25G which offers significant density, cost and power benefits for server to top-of-rack connections. Ever since the 25GbE was initially proposed in 2014, companies like Google, Microsoft, Arista, and Mellanox, etc. have been pushing the development of the 25GbE standard for top-of-rack server networking. Subsequently, the hubbub about 25GbE continues to rise and rolls out rapidly on the market, delivering the industry’s most comprehensive host and storage connectivity solution. Figure 1 highlights some of the key milestones from the year of 2014 to 2018 that have led to the creation of 25 Gigabit Ethernet in the industry.

Evolutionary Path of 25G Ethernet Figure 1: Evolutionary Path of 25G Ethernet

Migration Paths Comparison: 10G - 40G vs 25G - 100G

Before the 25GbE specification was released, the 10G - 40G upgrade path is predominantly adopted as an option for enterprises, service providers and data centers to scale beyond 10GbE. However, with the emergence of 25GbE, the 25G - 100G upgrade path has gained more momentum by offering the cost per bit, power consumption and server rack density advantages which are the necessary enablers for widespread speed transition. Let’s take a look at how they differ from each other and which migration path is more superior than the other one.

10G - 40G vs 25G - 100G Figure 2: 10G - 40G vs 25G - 100G Upgrade Path Comparison

10G vs 25G

In current data centers, the networking has reached a crossroad between 10G vs 25G for server-to-access switch connectivity. The 25G Ethernet based on the SFP28 form factor delivers 2.5 times more performance and bandwidth compared to 10G speeds, as well as supporting technology advancements from 10G in packaging and silicon. These benefits allow existing 25Gb switch architectures to support link speeds faster than 10G with no increase in cable/ trace interconnect, while keeping pace with the growth trajectory of networking bandwidth becoming faster and richer.

Besides, upgrading from 10GbE to 25GbE speeds offers both CapEx (capital expenditures) and OpEx (operational expenditures) savings for data center operators through backward compatibility. Recent high performance 25G/100G chips use single-lane 25G serdes technology similar in operation to 10GbE, thus reducing the power and cost per gigabit significantly. This power savings will in turn result in lower cooling requirements and operational expenditure for data center operators. By reusing the existing cabling infrastructure, the 25G Ethernet enables seamless network migration to avoid costly and complex changes.

 Can the SFP28 be used in SFP+ slot, and what speed will I get?

Yes, the pinouts of SFP28 and SFP+ are mating compatible. But SFP+ is designed to operate at 10Gb/s while SFP28 is designed to operate at 25Gb/s. That is to say, plugging an SFP+ into an SFP28 port would not get you 25Gb/s data rates but only 10Gb/s. Theoretically, plugging an SFP28 module into the 10G interface is feasible for certain devices to get 10Gb/s data rate, but this solution is not recommended, because it would be limited by the NIC and switch port that you have.

 Can the QSFP28 port generally accommodate the SFP28 or SFP+ optics?

The QSFP28 (4x25G lanes) can accommodate the SFP28 (25G lane) directly, but the QSFP28 cannot break out into 10G SFP+ links directly. If you want to connect 10G optics to the QSFP28 port, you'd need to insert the QSFP+ module (4x10G lanes) on the QSFP28 port, then use the breakout cable to convert the MPO to 4x LC and connect your standard 10G SFP+ modules on the far ends.

25G vs 40G

The 25G Ethernet specification places interoperability between a server NIC and top of rack (ToR) switch. It provides greater port density with maximum switch I/O performance and fabric capability compared to 40GbE solutions. 25GbE technology delivers 4 times the switch port density of 40GbE (four lanes) by requiring just one lane, enabling the network bandwidth to be effectively scaled in cloud and web-scale data center environments. The 25Gb/s and 50Gb/s Ethernet links provide 2.5 times faster performance per serdes lane and twinax copper wire than existing 10Gb/s or 40Gb/s Ethernet connections.

For any large and high-end enterprise, the port density per server will largely determine the cabling and switch infrastructure cost in the whole system. Therefore, upgrading directly from 40G to the faster 100G is cost prohibitive compared to the 25G Ethernet connectivity. The 25G - 100G (4x25G lanes) networking migration path provides a lower cost per unit of bandwidth by fully utilizing switch port capabilities when compared to 40G - 100G upgrade path.

 Do I Have to Move the Existing 40G Network to 25G?

Considering the 25G equipment is generally more expensive than 40G equipment, do I have to move the existing 40G network to 25G? It depends. 25G is definitely the way to go vs 10G/40G, since the cost per 25G lane is actually pretty reasonable. If you have a demand for increased baud rates or are planning for further upgrade to higher speeds (100G/200G/400G), it is imperative to deploy 25G as the roadmap to do the migration. If not, you don’t have to move the existing 40G network to 25G.

25G - 50G - 100G - 200G - 400G Market Highlights

25G servers and 100G switch ports have become ubiquitous in hyperscale data centers, gradually replacing previous 10G servers and 40G switches. This speed migration has boosted overall system throughput by 2.5x with small incremental costs. As the Ethernet industry continues to innovate and lay a path to higher networking speeds like 200G and 400G, the 25G - 100G upgrade has been developed as an inevitable roadmap for future data centers.

50G Is the New 40G - Why Do We Need 50G?

Although 50Gbps single-channel technology has been proposed to serve as the foundation for 100G/200G/400G upgrade, the implementation of the IEEE standard for 50G Ethernet is still some time away. Just like 25G delivers 2.5 times performance of 10G, the future 50G will also deliver 1.25 times performance of 40G. Since the 50G Ethernet will be based on using two 25G lanes, it will become an alternative to the current 40G implementations which use four 10G lanes to transport the Ethernet signal. The reduction of lanes will lower the cost of the network equipment needed in the data centers. In the future, the Ethernet upgrade path may change to 10G -> 25G -> 50G -> 100G from the traditional 10G -> 40G -> 100G. In any case it will be easy and cost effective to upgrade data centers from multiple 25G lanes to 50G or 100G networks.

Current vs Future High Speed Ethernet Upgrade Mode Figure 3: Current vs Future High Speed Ethernet Upgrade Mode

Beyond 100G, Upgrade to 200G and 400G

Networking infrastructures with 25G, 50G and 100G capabilities are typically implemented as very flexible solutions to pave the way for further upgrades to 200G and 400G in a large scale. This shift is now underway in high-end enterprises and data centers, driving forces primarily to the implementation of networks within the mega data centers and the interconnects between the data centers. More and more vendors on the market today have devoted themselves to the research and develop of 200G/400G optics, some of which have already been put into use successfully. The underlying signaling rates will range from 25G to 50G and perhaps ultimately land on 100G to support 200G/400G transmission. The following chart lists the feasible upgrade paths of 200G/400G based on multiple-lanes 25G/50G/100G. As you can see, 4 x 50Gb/s lanes or 8 x 25Gb/s lanes will be used to support 200G, while the higher 400G is expected to be achieved by 16 x 25Gb/s lanes, 8 x 50Gb/s lanes or 4 x 100Gb/s lanes.

200GbE Standard Lanes 400GbE Standard Lanes
200GBASE-SR4 4 x 50Gb/s 400GBASE-SR8 8 x 50Gb/s
200GBASE-SR8 8 x 25Gb/s 400GBASE-SR16 16 x 25Gb/s
200GBASE-DR4 4 x 50Gb/s 400GBASE-DR4 4 x 100Gb/s
200GBASE-FR4 4 x 50Gb/s 400GBASE-FR8 8 x 50Gb/s
200GBASE-LR4 4 x 50Gb/s 400GBASE-LR8 8 x 50Gb/s

Conclusion

The demand for higher Ethernet speed and performance never stops for future data centers. After looking back the evolutionary path of 25G in the past few years, it is easy to find that the emergence of 25G is a milestone to catch the leading edge broadband network capacity for next-generation data center demands. By offering more bandwidth and higher port density with reduced power consumption and cost, the 25G - 100G upgrade path overturns the traditional 10G - 40G connectivity to improve data center efficiency, thus laying the foundation for further upgrade to 200G and 400G. Let’s wait and see how this evolution will take place and hit the data center in continuous innovative methods.


    Facebook LinkedIn Twitter

DC Cabling Solution Team
DCI Engineers w/CCIE/HCIE
Certification

Ask Our Experts

*Free CCIE/HCIE Support

*Network Design/ Redesign, Analysis

*One-to-one service for each online PO

Thanks for your feedback.

Your opinion matters to us. We'll review and use it to improve the FS website for future visits.

Share Feedback

Thanks for visiting FS. Your feedback will help us provide customers a better experience.

How likely are you to recommend FS to your friends?*

1
2
3
4
5
Not likely Extremely Likely

Please select a topic for your feedback.*

0/1000

If you would like to hear back from us, please leave your contact information.