The Deployment of 25 Gigabit Ethernet in Data Center

Although 10 GbE (Gigabit Ethernet) to 40/100 GbE migration has been widely recognized as the Ethernet speed upgrading path, users still keep pursuing a better solution to replace the existing “10GbE-40GbE-100GbE” path. Companies like Google, Microsoft, Arista, and Mellanox are pushing the development of a 25 Gigabit Ethernet standard for top-of-tack server networking. Some may question the need for this technology, but they will soon see the benefits. New deployments, such as 10GbE—25GbE—100GbE or 10GbE—25GbE—50GbE—100GbE are announced to better serve the data center and cloud network. Thus, what is 25 GbE and why is it in demand?

What Is 25 GbE?

25GbE (25 Gigabit Ethernet) is a proposed standard for Ethernet connectivity in a data center environment. An industry consortium (25G Ethernet Consortium) was formed in July 2014 to support the specification of single-lane 25Gb Ethernet technology, because the proposed 25 GbE standard will use the same physical silicon from a single 25 Gbit/s lane. This simplifies the process with just minor changes for forward error correction and lane alignment, and it reduces the cost when compared to 40 GbE.

Why Is 25 GbE in Demand?

As is known to all, in the high-density data center, using multiple 10 GbE would require twice as many Ethernet switches with their associated space, power, and cooling costs. Deploying 25GbE networks enables organizations to significantly decrease capital and operating expenses by reducing the required number of switches and cables to solve these issues, compared to 10GbE and 40GbE (4×10 GbE) technology. Additionally, fewer network components also reduce ongoing management and maintenance costs.

For instance, if we use the 10GbE—40GbE—100GbE path, we will have 10 GbE single, 40 GbE quads and 100 GbE ten lanes in production. But when we turn to 25 GbE, we just need 25 GbE single, 50 GbE dual and 100 GbE quads in production. Obviously, 25GbE enables us to have 2.5X the performance of 10Gb Ethernet, making it a cost-effective upgrade to the 10GbE infrastructure. While compared to 40GbE, which is actually four 10GbE lanes, 25GbE is delivered across a single lane which provides greater switch port density and network scalability. Moreover, using multiple 25GbE lanes, it is easy to upgrade of 50GbE and 100GbE networks. It is a cost-effective solution for datacenter upgrade and cloud-scale network expansion. This is why 25 GbE is favoured and highly recommended by those famous consortium.

SFP28 & QSFP28 Assemblies

The SFP28 (25G Small Form-Factor Pluggable) and QSFP28 (25G Quad Small Form-Factor Pluggable) transceivers and interconnect cables are high-density, high-speed product solution designed for 25GbE and 100GbE applications in the telecommunications, data center and cloud-scale networks. The emergence of these two form-factors pluggable certainly reflect the trend in the industry to aggressively bring 100GE density up and costs down.

Based on the SFP+ MSA form-factor, SFP28 assembly solution enables a new generation of high-density 25G Ethernet switches and NIC cards, facilitating server connectivity in data centers, and a conventional and cost-effective upgrade path for enterprises deploying 10G Ethernet links today in the ubiquitous SFP+ form factor.

QSFP28 transceiver, as a new type of 100G transceivers, offers four channels of high-speed differential signals with data rates ranging from 25 Gbps up to potentially 40 Gbps, and will meet 100 Gbps Ethernet (4×25 Gbps) and 100 Gbps 4X InfiniBand Enhanced Data Rate (EDR) requirements. According to IEEE 802.3bm, the 100GBASE-SR4 QSFP28 is designed for multimode application and support maximum link length of 100 m over OM4 Fiber. The 100G QSFP28 LR4 module is designed for single-mode application which support maximum link length of 10 km over SMF. QSFP28 has the same footprint and faceplate density as QSFP+ and is just slightly smaller than CFP4. Theoretically, QSFP28 seems to have the density advantage over CFP4, but CFP4’s higher maximum power consumption gives it the advantage on longer reach optical distances. As the two main types of 100GbE transceivers, each of them has its own merits. Only time will tell how this all plays out.

Conclusion

Through the above analysis, we can see that, 25 GbE solution is more suitable for the high-density data center. But at present, for long distance transmission, the existing 40/100GbE solution—QSFP/QSFP+ and CFP family (CFP, CFP2, CFP4) seems to be better. FS.COM offers a comprehensive solution of fiber optic transceivers and cable assemblies. For data center, we offer a full product line of basic transceiver optics, such as 1000BASE-SX, 1000BASE-LX/LH SFPs, 10GBASE-LR SFP+ etc. We also offer high-density interconnection solution by launching whole series of 40GBASE QSFP+ optics and 100GBASE-LR4 CFP2 and CFP4 optics as well as the cable assemblies.

Introduction to XFP Transceivers

As telecom technology advances in the past few decades, the size of transceiver modules has become smaller and smaller. Meanwhile, the rate of optical transceiver also becomes faster and faster, for it is ranging from 1.25 G, 2.5 G, 4 G to 10G, 40G and 100G, etc. However, when it comes to 10G, people found that the encapsulation is too small to keep many components. Thus, XFP modules were launched in 2002 as a new standard to solve this problem. Next, this article will introduce the 10G XFP optical transceivers in great details.

Introduction of XFP Transceiver

In fact, XFP is not the first 10G transceiver in the market. The 10G transceiver market has gone through 300Pin and XENPAK / XPAK / X2 before XFP came out. As these types of products are integrated with SerDes chips, we usually call them transponders instead of transceivers.

XFP optical transceivers support up to 16 channels, which can plug into a standard server equipment rack. Although high port density is necessary to reducing costs, addressing heat dissipation has become a critical issue as well. Power requirements can be met using VCSEL technology and also using the latest silicon. XFP 10 Gbps optical transceivers have greatly accelerated and converged in terms of development efforts.

XFP optical transceivers have been designed to conform to the requirements of servers and to the performance and distance requirements of 10 Gbps Ethernet. The desired server transmission rates and encoding qualifications can easily be met by an XFP optical transceiver. As server technology evolves, optical transceivers follow its lead accordingly. When designers moved digital functions out of transceivers and into ASICs (application-specific integrated circuits), the efficiency of the transceivers improved.

Inefficiencies of XFP Transceivers

Overcoming the loss associated with optical transceivers is one of the primary concerns with XFP optical transceivers. Crosstalk is another concern. As technology improves and inefficiencies are minimized, servers will become more efficient. This effort requires some assistance from designers and manufacturers to aid in the evolution of servers and transceiver technology.

Keep in mind that optical transceivers in a server convert electrical signals into optical signals. This is what allows them to achieve speeds in excess of 25 Gbps. If they are bulky, however, they are more prone to transmission losses and signal degradation-another impediment to faster transmission.

To address these issues, circuit technology has emerged in conjunction with XFP transceivers to suppress reflections and increase data rates. Efforts have also been made to decrease costs and footprint by incorporating flexible printed-circuit boards and optical devices. Some companies have developed film-type lens sheets to address these issues, but they have not eliminated them entirely. Future technology may erase this impediment from existence.

XFP can increase port density and will reduce the costs of optical transceivers. For that reason, all designers should have a plan to upgrade their transceiver technology and server technology. With plans in place, business owners can expect to decrease processing time and also conserve space . Keep track of transceiver technology advancements and determine how it can help you achieve your long and short term goals.

Developed Trend

By 2006, with the pre-emphasis circuit and the equilibrium technology matured, it is no longer difficult to transmit signals at the rate of 10Gbps on a single fiber on the host board. In consequence, the SerDes chip has been no longer needed since then . Additionally, the smaller optical components and higher IC integration have made the miniaturization become possible. 3rd generation transceivers, namely XFP and SFP +, emerged in the market and still in wide use in today's 10G application.

As the saying goes, to want fish on the hook, the first thing you should know is what fish likes. When it comes to transceiver, it's the same situation. The developed trend of transceiver depends on what customers need. And the fact is that customers always need faster, smaller, lower power consumption, more functional, and more cost-effective transceivers. Different package of transceivers have come to the market one wave after another. With the fact that the new generation of transceiver package form will gradually replace the older generation, the old version also need to upgrade otherwise it has to be replaced gradually till disappear. For example, when XFP came to us, the transponder was replaced gradually. By adding tunable technique, the 10G transponders have still been in the market so Far. But Now SFP Tasu Module (Like HP J9151A ) Also Achieves Tunable Technique Which Indicates That XFP Will Vanish Sooner Or Later.

1000Base-T SFP Module for Gigabit Ethernet

The Gigabit Ethernet technology is an extension of the 10/100-Mbps Ethernet standard. Gigabit Ethernet provides a raw data bandwidth of 1000 Mbps while maintaining full compatibility with the installed base of over 70 million Ethernet nodes. Gigabit Ethernet includes both full- and half-duplex operating modes. A Gigabit Ethernet is imperative for two reasons: faster systems and faster backbones. Gigabit Ethernet has the potential for low-cost products, freedom of choice in selecting the products, interoperability, and backward compatibility. Gigabit Ethernet supports existing applications, network operating systems, and network management; it requires a minimal learning curve for Ethernet network administrators and users. These investment preservation and risk minimization aspects are what make Gigabit Ethernet so attractive. With the development of Ethernet systems and the growing capacity of modern silicon technology, embedded communication networks are playing an increasingly important role in embedded and safety critical systems.

A known type of data communication device is a small form factor pluggable (SFP) module. Typically, the SFP module plugs into an interface slot in a circuit board populated with other communication devices used in an Ethernet-based system. The SFP module includes a second serial interface,interconnected with the circuit board slot, and a first serial interface, coupled to a serial link, such as a copper or fiber link, for communicating with remote link partners. The serial link, coupled with the first serial interface, may be a 10/100/1000 Base-T copper link, or a fiber link, for example. The SFP module also offers several significant advantages over its predecessor, the GBIC (Gigabit Interface Converter), including lower cost, lower power, and smaller size. Thus, with the SFP form factor, fiber Gigabit systems may be developed featuring similar port densities as copper-only systems using RJ-45 connectors.

The SFP transceiver MultiSource Agreements (MSA) document puts forward a specification for the development of optical SFP modules based on IEEE 802.3z, the Gigabit Ethernet Standard. Specifically, the MSA calls out 1000Base-X Physical Coding Sub-layer (PCS) which supports full-duplex binary transmission at 1.25 Gbps over two copper wire-pair SerDes (Serializer/Deserializer). Transmission coding is based on the ANSI Fiber Channel 8B/10B encoding scheme.

1000Base-X makes no provision for running at slower speeds. Thus, network device ports utilizing SFPs are dedicated to operating on fiber links at speeds of 1000 Mbps. However, more than 85% of office space inside buildings is category 5 copper. Thus, ports designed to use optical SFPs can not make use of this existing cabling.

For example, a customer may require a network device, such as a router, having both optical ports for long distance connections and RJ-45 copper ports for connecting to local devices. It is often the case that not all optical ports provided on a router are needed at a given time. However, with conventional SFPs these optical ports cannot be utilized to connect with local devices connected by standard copper cabling or operating at speeds lower than 1000 Mbps. But with a 1000BASE-T copper SFP transceiver, the customer could use their existing copper cabling infrastructure instead of replacing the infrastructure. Here are two good examples of 1000BASE-T copper SFP transceivers, the Finisar FCLF-8521-3 compatible 1000BASE-T SFP copper transceiver and Cisco Linksys MGBT1 compatible 1000BASE-T SFP copper transceiver from FS.COM. Both of them are designed for 100m reach over Cat 5 UTP cable with RJ-45 interface and support max data rate of 1000Mbps.

The 1000BASE-T copper SFP transceiver offers a flexible and simple method to be installed into SFP MSA compliant ports at any time with no interruption of the host equipment operation. It enables for seamless integration of fiber with copper LAN connections wherever SFP interface slots can be found. Such system is economical, it saves time, offers flexibility and eliminates the necessity for replacing entire devices once the customers have to change or upgrade fiber connections and you will benefit so much from it.

Interconnect Trends in the Data Center

In order to better satisfy the demands for high speed, wide bandwidth and high density, changes are taking place in the data center. Driven by the cloud services, the hyperscale/cloud data centers are becoming larger, more modular, more homogeneous with an architecture change from traditional 3-tier to flattened 2-tier topology. To deal with these changes and challenges, what will we need for optical interconnection a few years down the road? Here are the key trends to watch in 2016 which give you some guides to the roadmap.

Significant Increase in 100G & 25G Port Density

For easy understanding, we divide the data center into three parts—intra-rack, inter-rack and long span/inter-building. Within the data center rack, 10 GbE is being widely deployed now but 25 GbE will be deployed soon and 50 GbE to the server will follow. Between data center racks, 40 GbE now is deployed and 100 GbE or beyond will follow. When for long spans/inter data center and WAN (Wide Area Network), 100 GbE has been deployed until now and 400 GbE is standardized.

The overall trend of data center connection is moving from 10/40 G to 25/100 G. Small form factors, power dissipation under 3.5W, active optical cables (AOCs) etc., are becoming the trends. To satisfy the demands for high port density, the 100G optical module has gone through a revolution of form factor. Until now, QSFP28 is becoming the 100G module form factor of choice for new data center switches. QSFP28 is both a 100G and a high-density 25G form factor (4×25 Gbps). It will have very high volumes, because it supports both 100G and 25G links.

When compared to 40G cabling, we believe a 25G connection speed offers a more flexible and cost-efficient upgrade point from 10 GbE for cloud providers on the long-term path to 100G connections. As a 25G module form factor, SFP28 is the choice for new Servers / NICs (Network Interface Cards).

Extension of Optical Links Beyond the Standards

Duplex and parallel optics products of both 40 GbE and 100 GbE continue to proliferate which result in a proliferation of standards, de facto standards, MSAs (Multi-Source Agreements), and proprietary codes, each optimized for a particular use case. Meanwhile, various recent 25G and 100G Ethernet standards and MSAs require the use of RS-FEC (Reed-Solomon Forward Error Correction) on the host to increase overall link length. RS-FEC does not increase the total bit rate, but it introduces an additional latency of ~100ns in the link. As the fiber propagation time of each bit over 100m of MMF is ~500ns, the amount of additional latency introduced by RS-FEC may impact the overall performance of short links 500 meters. Thus, the low-Latency QSFP28 SR4 and SFP28 SR without FEC will be a trending option for data center short-reach interconnection.

Reutilization of Existing 10G Fiber Plant on 40/100G

One of the main interconnect trends in the data center is that data center operators want to upgrade from 10G to 40/100G without touching the duplex MMF (multimode fiber) infrastructure. Today’s data centers are deployed around 10 GbE, primarily focusing on 10GBASE-LR using LC duplex interface with SMF (single mode fiber). For instance, HP J9151A compatible 10GBASE-LR SFP+ transceivers from FS.COM are high-performance 10G optics using LC duplex interface to achieve max link length of 10 km over SMF. But in order to achieve 10G to 40G migration, the breakout cabling solution may be the preferred path for 40/100G migration as shown below.

The increased demand for blazing connection speeds and unlimited bandwidth drives the data center to change. The interconnect trends in data center market toward 25/100G with smaller module form factors for higher port density, lower power consumption and low cost per bit. Meanwhile, data center is required to increase performance to leverage existing fiber infrastructure that achieves 40/100G migration from 10 GbE. For higher speed beyond 100G, new Ethernet speeds including 50G, 200G, 400G are being standardized.

Reference: http://www.fs.com/blog/interconnect-trends-in-the-data-center.html

QSFP+ Direct Attach Copper Cables for EX Series Switches

Quad small form-factor pluggable plus (QSFP+) direct attach copper (DAC) cables are suitable for in-rack connections between QSFP+ ports of EX Series switches. They are suitable for short distances of up to 16.4 ft (5 m), making them ideal for highly cost-effective networking connectivity within a rack and between adjacent racks. This article will introduce EX Series switches and QSFP+ DAC for EX Series switches.

Introduction to EX Series Switches

EX Series switches deliver scalable port densities and carrier-proven high availability features that consolidate legacy switch layers, helping to reduce capital and operational expenses and advance the economics of networking. For example, the EX 4200 series Ethernet switches with Virtual-Chassis technology, deliver the same Gigabit Ethernet (GbE) and 10GbE port densities as traditional chassis-based switches, but at one-eighth the footprint and less than one third the cost. The EX Series switches are right-sized for campus, data center and remote office environments and feature many of the same carrier-class hardware and software architectures found in core routers that were purpose-built to support the convergence of data, voice, and video onto a single always-on network.

By alleviating the cost, complexity and risk associated with legacy switch infrastructures, the EX Series switches enable high-performance businesses to deploy a high-performance network infrastructure based on three key tenets – operational simplicity, carrier-class reliability, and integration and consolidation – to enable ubiquitous access to strategic assets, reduce network downtime and enhance overall security to shared assets across the extended enterprise.

Cable Specifications of QSFP+ DAC

QSFP+ direct attach copper (DAC) cable is hot-removable and hot-insertable. QSFP+ DAC mainly has two kinds. One is a cable that connects directly into two QSFP+ modules, one at each end of the cable. The cables use integrated duplex serial data links for bidirectional communication and are designed for data rates up to 40 Gbps. The other is a breakout cable consisting of a QSFP+ transceiver on one end and four SFP+ transceivers on the other end. The QSFP+ transceiver connects directly into the QSFP+ access port on the QFX Series device. The cables use high-performance integrated duplex serial data links for bidirectional communication on four links simultaneously. The SFP+ links are designed for data rates up to 10 Gbps each.

The following table describes the software support for QSFP+ passive DAC cable lengths on EX Series switches for Junos OS releases.

<colgroup> <col /> <col /> <col /></colgroup>

Switch	Software Support Added	DAC Model Number
EX44300 switches	Junos OS for EX Series switches, Release 13.2X51-D15 or later	EX-QSFP-40GE-DAC-50CM QFX-QSFP-DAC-1M QFX-QSFP-DAC-3M JNP-QSFP-DAC-5M
EX4300 switches	EX4300-24T, EX4300-24P, EX4300-48T, EX4300-48T-AFI, EX4300-48P, EX4300-48T-DC, and EX4300-48T-DC-AF switches—Junos OS for EX Series switches, Release 13.2X50-D10 or later EX4300-32F switches—Junos OS for EX Series switches, Release 13.2X51-D15 or later EX4300-24T-S, EX4300-24P-S, EX4300-32F-S, EX4300-48T-S, and EX4300-48P-S switches—Junos OS for EX Series switches, Release 13.2X51-D26 or later	EX-QSFP-40GE-DAC-50CM QFX-QSFP-DAC-1M QFX-QSFP-DAC-3M JNP-QSFP-DAC-5M
EX4550 switches	EX4550-32T-AFI, EX4550-32T-AFO, EX4550-32T-DC-AFI, EX4550-32T-DC-AFO, EX4550-32F-AFI, EX4550-32F-AFO, EX4550-32F-DC-AFI, and EX4550-32F-DC-AFO switches—Junos OS for EX Series switches, Release 13.2X50-D10 or later EX4550-32F-S switches—Junos OS for EX Series switches, Release 12.3R5 or later	EX-QSFP-40GE-DAC-50CM QFX-QSFP-DAC-1M QFX-QSFP-DAC-3M JNP-QSFP-DAC-5M

Conclusion

QSFP+ direct attach copper cables can provide cost-effective and reliable 40G speed connections for EX Series switches with distances reaching up to 10 meters. As the leading fiber optical manufacturer in China, FS.COM offers a wide selection of QSFP+ DAC with low cost but high performance. In addition, 10G SFP+ to SFP+ DAC (eg. HP JD096C), 25G SFP28 to SFP28 DAC, 40G QSFP+ to 4 XFP DAC, 100G QSFP28 to QSFP28 DAC, 100G QSFP28 to 4 SFP28 DAC are also available for your choice. All these DACs are with 100% compatibility and can be customized according to your special requirements.