Tag Archives: 10 Gigabit Ethernet

10G SFP or 10GBASE-T SFP for 10GbE Network?

The dramatic growth in data center requires the higher-performance servers, storage and interconnects. From initial 100M, 1G, 10G, to 40G and 100G, high speed Ethernet has never stopped developing. The standard for 10 Gigabit Ethernet (IEEE802.3ae) was ratified in 2002. In 10 Gigabit Ethernet, engineers often find it puzzled to choose a more suitable physical media between fiber and copper. Take a look at the media options for 10GbE Network.

Media Options for 10GbE Network

SFP+ (small form-factor pluggable plus) supports both fiber optic cables and DAC (direct attach cable). It delivers a wide variety of 10GbE Ethernet connectivity options for data center, enterprise wiring closet, and service provider transport applications. But it has the limitations that will prevent the media from moving to every server.

SFP+ cable is designed for 10GbE access layer interconnection in data center. It includes direct attach copper cables and active optical cables. DAC is a lower cost alternative to fiber, but it can support limited transmission distance and it’s not backward-compatible with existing GbE switches. DAC requires the purchase of an adapter card and requires a new top of rack (ToR) switch topology. DAC is more expensive than structured copper channels, and cannot be field terminated.

10GBASE-T SFP

10GBase-T SFP enables 10GbE connections with unshielded or shielded twisted pair cables over distances up to 100 meters. 10GBase-T technology appears as SPF is not compatible with twisted pair cabling system typically used in data centers. With 10GBase-T SFP, the migration from 1GbE to 10GbE can be easily achieved.

Comparison of 10G SFP and 10GBase-T SFP
Latency

Low latency becomes so important since the adoption of private cloud applications increases. It’s beneficial for ensuring fast response time and reducing CPU (center processing units) idle cycles so that improve data center efficiency.

As to 10GBASE-T SFP, the physical connection (PHY) standard uses block encoding to transport data across the cable without errors. The block encoding requires a block of data to be read into the transmitter PHY, a mathematical function run on the data before the encoded data are sent over the link. It happens the same on the receiver side. This standard specifies 2.6 microseconds for the transmit-receive pair, and the block size requires latency to be less than 2 microseconds. While 10G SFP applies simplified electronics without encoding, and common latency is around 300 nanoseconds per link.

You may think that two microseconds are not high. But what if a TOR infrastructure where traffic is passing 4 hops to reach the destination? 10.4-microsecond delay will be caused when using 10GBASE-T SFP. The following table tells details about the latency of SFP+ cable, 10G SFP and 10GABSE-T SFP for different number of links.

Number of Links SFP+ Cable Latency 10G SFP Latency 10GBASE-T SFP Latency
1 0.3 0.1 2.6
2 0.6 0.2 5.2
3 0.9 0.3 7.8
4 1.2 0.4 10.4
5 1.5 0.5 13.0
6 1.8 0.6 15.6

From the above table, it shows that the latency of 10GBASE-T SFP is the highest. As network links grow, the latency turns to be higher. It’s known that the lower latency, the faster the network speed. High latency in the data center infrastructure results in delays in CPU and application works, therefore limiting data center efficiency and increasing operational costs.

Power Consumption

Power consumption is also one of the important factors to be considered in data centers. Engineers are sensitive to power consumption and find a way to seek the lowest possible power consumption technologies. It’s said that every watt of power consumed, typically two additional watts are needed for cooling.

10GBase-T components today require anywhere from 2 to 5 watts per port at each end of the cable depending on the distance of the cable. But 10G SFP requires about 0.7 watt regardless of distance. The figure below compares the power consumption of three media options of 10GbE Ethernet.

10GBASE-T-SFP-power consumption

From this figure, suppose there are 10000 ports in the data center, 10G SFP can greatly save the power. On contrary, 10GBASE-T components consumes the most power. Thus, to save power in the data center, 10G SFP and SFP+ cable should better be selected when deploying thousands of cables in a data center.

Conclusion

From this article, 10G SFP and SFP+ cable solutions are better than 10GBASE-T SFP for 10G data center. But 10GbE is not the ultimate goal. Besides factors mentioned in this article, you should also select a cabling solution which can support not only current needs but also future data center deployments when you design 10GbE network. You can find various SFP+ modules and 40G QSFP+ from FS.COM.

Related article: How to Convert SFP+ to 10GBASE-T/RJ45?

Things You Need to Know Before Deploying 10 Gigabit Ethernet Network

Since the establishment of 10 Gigabit Ethernet, it has been employed by large amount of enterprises in their corporate backbones, data centers, and server farms to support high-bandwidth applications. But how to achieve a reliable, stable and cost-effective 10Gbps network? There are ten things you should know before doing the deployment.

More Efficient for the Server Edge

Many organizations try to optimize their data centers by seeking server virtualization which supports several applications and operating systems on a single server by defining multiple virtual machines on the server. Because of this, the organizations can reduce server inventory, better utilize servers, and mange resources more efficiently. Server virtualization relies heavily on networking and storage. Virtual machines require lot of storage. The network connectivity between servers and storage must be fast enough to avoid bottlenecks. And 10GbE can provide the fast connectivity for virtualized environments.

More Cost-effective for SAN

There are three types of storage in a network: direct-attached storage, network attached storage, and SAN. Among them, SAN is the most flexible and scalable solution for data center and high-density applications. But it costs much and needs special trainees for installing and maintaining the Fibre Channel interconnect fabric.

The internet small computer system interface (iSCSI) makes 10 Gigabit Ethernet an attractive interconnect fabric for SAN applications. iSCSI allows 10 Gigabit Ethernet infrastructure to be used as a SAN fabric which is more favorable compared with Fibre Channel. Because it can reduce equipment and management costs, enhance server management, improve disaster recovery and deliver excellent performance.

Reducing Bottlenecks for the Aggregation Layer

Today, traffic at the edge of the network has increased dramatically. Gigabit Ethernet to the desktop has become more popular since it becomes less expensive. More people adopt Gigabit Ethernet to the desktop, which increases the oversubscription ratios of the rest of the network. And that brings the bottleneck between large amounts of Gigabit traffic at the edge of the network and the aggregation layer or core.

10 Gigabit Ethernet allows the aggregation layer to scale to meet the increasing demands of users and applications. It can well solve the bottleneck for its three advantages. First, 10 Gigabit Ethernet link uses fewer fiber stands compared with Gigabit Ethernet aggregation. Second, 10 Gigabit Ethernet can support multi Gigabit streams. Third, 10 Gigabit Ethernet provides greater scalability, bringing a future-proof network.

Fiber Cabling Choices

To realize 10 Gigabit Ethernet network deployment, three important factors should be considered, including the type of fiber cable (MMF of MF), the type of 10 Gigabit Ethernet physical interface and optics module (XENPAK, X2, XFP and SFP+).

Cable Types Interface Max Distance
MMF (OM1/OM2/OM3) 10GBASE-SR 300 m
10GBASE-LRM 220 m
10GBASE-ER 40 km
SMF (9/125um fiber) 10GBASE-LR 10 km
10GBASE-ZR 80 km

Form factor options are interoperable when 10 Gigabit Ethernet physical interface type is the same on both ends of the fiber link. For example, 10GBASE-SR XFP on the left can be linked with one 10GBASE-SR SFP+ on the right. But 10GBASE-SR SFP+ can’t connect to one 10GBASE-LRM SFP+ at the other end of the link.

Copper Cabling Solutions

As copper cabling standards becomes mature, the copper cabling solutions for 10GbE is becoming common. Copper cabling is suitable for short distance connection. The are three different copper cabling solutions for 10 Gigabit Ethernet: 10GBASE-CX4, SFP+ DAC (direct attach cable) and 10GBASE-T.

10GBASE-CX4 is the first 10 Gigabit Ethernet standard. It’s economical and allowed for very low latency. But it’s a too large form factor for high density port counts in aggregation switches.

10G SFP+ DAC is a new copper solution for 10 Gigabit Ethernet. It has become the main choice for servers and storage devices in a rack because of its low latency, small connector and reasonable cost. It’s the best choice for short 10 Gigabit Ethernet network connection.

10GBASE-T runs 10G Ethernet over Cat6a and Cat7 up to 100 m. But this standard is not very popular since it needs technology improvements to reduce its cost, power consumption, and latency.

For Top of Rack Applications

A top-of-rack (ToR) switch is a switch with a low number of ports that sits at the very top or in the middle of a 19’’ telco rack in data centers. A ToR switch provides a simple, low-cost way to easily add more capacity to a network. It connects several servers and other network components such as storage together in a single rack.

ToR switch uses SFP+ to provide 10G network in an efficient 1U form factor. DAC makes rack cabling and termination easier. Each server and network storage device can be directly connected to the ToR switch, eliminating the need for intermediate patch panels. DAC is flexible for vertical cabling management within the rack architecture. And the cabling outside the rack, the ToR switch uplink connection to the aggregation layer, simplifies moving racks.

The following figure shows a 10 Gigabit Ethernet ToR switching solution for servers and network storage. Because the servers are virtualized, so the active-active server team can be distributed across two stacked witches. This can ensure physical redundancy for the servers while connected to the same logical switch. What’s more, failover protection can be offered if one physical link goes down.

10G-ToR

Conclusion

10 Gigabit Ethernet network is not the fastest but quite enough for common use in our daily life. So you should better read this article before you do the deployment. Besides, FS.COM provides both fiber and copper cabling solutions for 10G network. For more details, please visit www.fs.com.

The New 10G Multimode Optical Solution – 10GBASE-LRM

10 Gigabit Ethernet has been applied for a long time in data centers and enterprise LANs. For 10G Ethernet connection, there are both single-mode and multimode solutions. First let’s see the original multimode solutions and supportable distances for 10G Ethernet.

10G-multimode solution-supportable-distance

10GBASE-S operates at 850nm wavelength. It can support up to 300m distance over laser-optimized OM3. This makes it a popular standard for data centers and cooperate backbones. For the conventional OM1 and OM2 which are not optimized for laser transmission, the furthest supportable distance is 33 m and 82 m. So these two solutions are only used in equipment rooms or small data centers.

10GBASE-LX4 was specified to support 300 m over three cable types. So it relies on coarse wavelength division multiplexing (CWDM) which is more complex and expensive technology. 10GBASE-LX4 operates at 1300nm wavelength and that requires additional cost on mode-conditioning patch cords (MCPCs).

The high cost and relatively slow adoption of 10GBASE-LX4 drive the development of a new standard—10GBASE-LRM. 10GBASE-LRM is developed to offer a longer reach for conventional fiber cables at a lower cost and smaller size than 10GBASE-LX4. The following will talk about 10GBASE-LRM from three sides.

Transmission Distance

On condition the supporting distance, 10GBASE-LRM can only support 220 m. It’s suitable for LAN networks within buildings. But a cabling survey provides that for 10G network, the distance is not able to address 30% of in-building channels.

Electronic Dispersion Compensation

The key to the long reach of 10GBASE-LRM on conventional multimode fiber is electronic dispersion compensation (EDC). EDC is deployed as an integrated circuit that acts like a complex filter on the received signal from the optical fiber. The purpose is to extend the maximum supportable distance. 10GBASE-LRM applies EDC technology and is therefore independent of the optical wavelength. 10GBASE-LRM operates at 1300 nm.

EDC chips is added to a linear detector in the receiver. As an additional component, it increases cost, consumes power and wastes heat. It can only work as intended in conjunction with a linear detector and amplifier. Because the EDC device must operate on a faithful analog rendition of the optical waveform in the fiber. For 10GBASE-LRM, to reproduce the optical waveform with precision, extra requirements and cost on the receiver design are needed.

Multiple Transmit Launch Conditions

In order to improve the chances of operating at a higher bandwidth, 10GBASE-LRM relies on multiple transmit launch conditions.

One launch is achieved by using mode-conditioning patch cord. The other launch is produced using a regular multimode patch cord. Through the two launches, different modes can be achieved and a favorable operating condition can be easily found.

There are four possible patch cord combinations at both ends of the channel. The preferred launch uses MCPCs on both ends. This process requires a test for link stability for each configuration. The user should shake and bend the patch cord at the transmit end while observing channel health indicators at the receive end. The shaking and bending of the cords causes changes to the received waveform which the receiver must tolerate in normal operation. If there were transmission errors, then users should change another launch. The errors indicate that the channel is operating near or beyond the limit of the receiver’s capability and the link may fail in operation.

launch-conditions-for-10gbase-lrm

However, the 10GBASE-LRM standard’s committee refuse to implement this channel test. So the burden of the shaking and bending lies on the users. It’s not good for the popularity of 10GBASE-LRM.

Comparison of Several 10G Transceivers Cost

The following will compare the cost of 10G transceivers from several sides, including laser, receiver, package and cords.

Laser: 10GBASE-LRM uses 1310nm fabry perot lasers, which cost fewer than 10GBASE-L’s and 10GBASE-LX4 DFB lasers, but more than 10GBASE-S’s 850nm VCSELs. 10GBASE-LRM requires tighter transmitter waveform control to limit the transmit waveform dispersion penalty that EDC can’t compensate. Thus, it reduces transmitter yields and increases cost.

Receiver: 10GBASE-LRM adds EDC chip cost to receiver and needs a linear detector and amplifier instead of other cheap digital equipment.

Package: 10GBASE-LRM requires a smaller package than 10GBASE-LX4. However, not like 10GBASE-S, 10GBASE-LRM requires higher-cost single-mode transmitter alignment for compatibility with mode conditioning patch cords.

Cords: 10GBASE-LRM needs mode conditioning patch cords for reliable link operation. And the cost is much higher than regular SMF or MMF fiber optic patch cords.

Through the comparison among these 10G optical transceivers, you may find which one costs fewer. 10GBASE-LRM transceiver is cheaper than 10GBASE-LX4, more expensive than 10GBASE-L and 10GBASE-S transceivers.

Conclusion

10GBASE-LRM is a multimode solution for 10 Gigabit Ethernet. Based on the above content, 10GBASE-LRM has some advantages over 10GBASE-LX4. It offers lower cost and smaller package. But the distance and reliability are not very ideal. Compared with 10GBASE-S, 10GBASE-LRM is not so good as to the cost, simplicity, reliability and distance capability. FS.COM provides various types of cost-effective 10GBASE transceivers, such as 10GBASE-LR, 10GBASE-SR, 10GBASE-ER, etc. Other compatible brands like Cisco, Juniper, Arista, Brocade are also available. Among so many choices, you must choose the most suitable solution for your network connection.

Basic Knowledge about Ethernet

Nowadays, billions of file cabinets and mountains of papers stored in computers need to be transmitted at high speed with great efficiency. Computer networking technologies are key to meet this demand, allowing computers on the internet to send and receive information easily. This article will introduce the network technology: Ethernet which is widely used nowadays.

Ethernet is a family of computer networking technologies for local area networks (LANs) and metropolitan area networks (MANs), connecting more than 85 percent of the world’s LAN connected PCs and workstations. It is a link layer protocol in the TCP/IP stack, describing how networked devices transmit data on the same network segment and how to put data out on the network connection. Ethernet was commercially released in 1980 and first standardized in 1983 as IEEE 802.3. Its standards have been updated to embrace new media, higher transmission speeds and changes in frame content such as the new standard 802.3af defining Power Over Ethernet [POE] crucial to most Wi-Fi and IP telephony deployments.

Ethernet

Ethernet was initially designed to run over coaxial cables but has been updated to used for twisted pair cables and fiber optical fibers over years. The most commonly installed Ethernet systems are called 100 BASE-T (the “BASE-T” part means the systems use twisted-pair cabling) which provides transmission speeds up to 100Mbps. It is typically used for LAN backbone systems, supporting workstations with 10BASE-T cards. Another widely used one is Gigabit Ethernet which is primarily carried on optical fiber with very short distances possible on copper media. It provides an even higher level of backbone support at 1000 Mbps or 1 Gbps. With the increasing of data transfer rates, the standards: 10 Gigabit Ethernet and 100 Gigabit Ethernet are available. Their data rates reached up to 10 gigabits per second and even 100 gigabits per second respectively, making them be good solutions to deliver high bandwidth in LANs.

QQ截图20150525174629

100GBASE-LR4 CFP2 Optical Transceiver Module

It is concluded that Ethernet has evolved to provide excellent performance and network intelligence. As Ethernet data transfer rates are excepted to be increased to 400 Gbit/s by early 2017, Ethernet will be the most potential network technology in the future for its high-speed data transmission.

Fiberstore offers a wide range of products for 10GbE or 100 GbE applications such as the new product: 100GBASE-LR4 CFP2 Optical Transceiver module. The optical transceiver offers smaller size and lower power consumption for data center networking, enterprise core aggregation, and service provider transport applications. For more information, please visit www.fs.com.