May 2017

25G Cabling Solutions: SFP28 Transceiver VS. SFP28 Twinax Cable VS. Cat8 Cable

25. may 2017 at 6:18 | jack
Developed by IEEE 802.3 task force P802.3by, 25G Ethernet is a new standard for Ethernet connectivity in a data center environment. This standard was derived from 100G Ethernet standard, however, its operation works as a single lane connection with 25Gbps that can be run on fibers or coppers. To support 25G Ethernet, there are three most common cable solutions: SFP28 optical transceiver, SFP28 twinax cable and cat8 cable, which one is better for your 25G network? Please continue read the post, and you will find the answer.

SFP28 Transceiver
Fiber optic module and fiber patch cable are always the first choice in network connectivity. In this connection method, two things will be required: 25G SFP28 modules and duplex LC optical jumper. You need firstly plug the patch cable into SFP28 modules, then plug the whole assembly into the both devices. There are main two types of SFP28 transceivers used to transmit data across fiber optic cables with different maximum distances as shown in the following table.

SFP28 standard
SFP28 Transceiver

Pros: SFP28 module and fiber cable is the best choice when the distance between two data center servers is very long.

Cons: Fiber and module is often more expensive than other two cabling solutions.

SFP28 Twinax Cable
Intended for short runs (up to 5 meters), SFP28 twinax cable or SFP28 direct attach twinax cable (DAC) is a cable assembly which is terminated with two SFP28 connectors at both ends. As the other DAC cables (eg: SFP+ DAC or QSFP+ DAC) that we are more familiar with, the SFP28 connector is also not real transceiver module, which doesn't have expensive optical lasers, thus making this solutions more cost-effective than SFP28 transceiver solution. SFP28 twinax cable is ideal for high density, high speed I/O data center applications in the networking, telecom and data storage markets where maximum overall network efficiency and lower overall cost are desired.

SFP28 DAC cable

Pros: It provides an extremely efficient increase in speed to top-of-server (ToR), and it is very suitable for very short links and offer a highly cost-effective way to establish a 25-Gigabit link between SFP28 ports of switches within racks and across adjacent racks.

Cons: Transmission distance is usually less than 10 meters.

25GBase-T: Cat8 Cable
While 25G over twinax DAC assemblies will fulfill the ToR server environment where a distance of 3 to 5 meters is more than adequate, there is also the need for longer distances to support the middle of the row (MoR) topologies to about 15m and end of row (EoR) to 30m. That's where a 25GBase-T application over balanced twisted-pair copper cabling has potential to fill the gap. The Cat8 cabling is designed to support emerging 25GBase-T and 40GBase-T applications, which are specified over an extended bandwidth of 2GHz for a distance over 30 meters. It is fully backward compatible with Cat6a cabling, including RJ45 connectivity, and supports all Cat6a applications such as 10GBase-T for a distance of 100 meters.

cat8 cable

Pros: Longer transmission distance than SFP28 twinax cable, up to 30 meters. Fully backward compatible with Cat6a cabling, which will provide easier migration from 1G to 10G to 25G.

Cons: The technology of 25GBase-T hasn't been very mature yet, so there is not many devices that can support 25GBase-T standard.

Taking the advantages of high density and low cost consumption with no changes in architecture required, there is no doubt that 25G Ethernet will have a broad market potential in the server interconnect world. In the previous text, we have introduced three cabling solutions: 25G Ethernet over fiber, twinax and fiber. You can choose the right solution according to your demands.

Understanding Array Polarity With Parallel Link

15. may 2017 at 10:51 | jack
The use of pre-terminated fiber assemblies and cassettes is growing, and the deployment of systems with speeds up to and beyond 100G are on the horizon for many users. As a result, the issue of maintaining polarity in parallel fiber-optic links is becoming increasingly important. In the previous posts, we have introduced polarity in point-to-point duplex links which is achieved through what is known as an A-to-B patch cord. In this post, we are going to talk about array polarity with parallel link.

Array Polarity With Parallel Link Overview
Array polarity with parallel link has the corresponding Method A, B and C to establish polarity for parallel signals using an MPO transceiver interface with one row of fibers. For example, 40 Gigabit Ethernet over multimode fiber uses 4 transmit and 4 receive fibers in a 12-fiber array, or 4 lanes at 10Gbps. In order to understand these polarity methods more specifically, we can make a comparison with polarity methods for duplex signals. From the following table, we can easily find out that the breakout MTP cassette and the duplex fiber patch cords in duplex link are replaced with 12-fiber array patch cords that plug directly into the MTP adapter at the patch panel and into the equipment interface in parallel link.

polarity of multiple duplex signals vs. parallel signals
Three Methods for Array Polarity With Parallel Link
Method A
Method A as shown below recognized in 568-C.0 uses Type A backbone on each end connected to a patch panel. On one end of the optical link, a Type A array patch cord is used to connect patch panel, while on the other end, a Type B array patch cord is used to connect patch ports to their respective parallel transceiver ports.

Method B
Also recognized in 568-C.0 uses Type B throughout-Type B array cable, Type B adapters and Type B array patch cords to achieve the whole optical link. More detailed information can be seen in the following image.

Method C
The proposed Method C as shown in the image below is similar to Method A, but it would use Type C trunk cable instead of Type A, and a Type C cross-over patch cord is required at one end and the other end uses Type B patch cable.


Note: An important point to remember is that MPO plugs use alignment pins. For a MPO connection, one plug is pinned and the other plug is unpinned. As MPO transceiver typically has pins, this convention leads to the following implementation on initial build out: 1) Patch cords from transceiver to patch panel are typically unpinned (female) on both ends. 2) Trunk cables are typically pinned (male) on both ends.

As duplex link, there are also three methods for parallel link. However, maintaining array polarity with parallel link is not as simple as it seems. This article can only provide some basic information about the polarity with parallel link. In the following updating, we will talk about more about array polarity system.

Efficiently and Conveniently Integrate 10G, 40G, and 100G Equipment With MTP Breakout Patch Panel

11. may 2017 at 12:17 | jack
Not all that long ago, Ethernet networks that supported 10G speeds were considered amazingly fast, and now 40G is the norm in most data center. As 40G Ethernet becomes a standard in data centers, the challenge of connecting 40G equipment with existing 10G equipment moves front and center. Adding further complexity, it's clear that organizations of all sizes also need to be prepared to integrate speeds of 100/120G and beyond. MTP breakout cable filled a pressing need when no other options were available, but using unstructured cabling makes installs, upgrades, changes, troubleshooting and repairs extremely inefficient. The emergence of MTP breakout patch panel allows you to seamlessly and conveniently integrate equipment with different network speeds to meet your connectivity need today.

The Nuts and Bolts of MTP Breakout Patch Panel
MTP breakout patch panel is integrated with a range of modular, removable fiber cassettes in a rack mount patch panel, which combines the functionality of breakout cables, the efficiency of structured cabling and the convenience of a pre-assembled kit. Breakout patch panel splits 40G QSFP+ and 100G CFP switch port into 10G duplex LC ports, which connect to devices' SFP+ ports with high-quality off-the-shelf fiber patch cables.

working principle of MTP breakout patch panel
Sparkles of MTP Breakout Patch Panel
Convenience and Efficiency: Pre-assembled panels including modular fiber breakout cassettes with build-in MTP cable and duplex LC ports makes it possible for quicker deployment. In addition, its structured cabling makes installs, upgrades, changes, troubleshooting and repairs quicker, easier and more cost-effective than using MTP breakout cables.
Space Saving: By managing varying port densities and speeds in a single high density patch panel, you save valuable rack space, helping to lower data center costs. A 1U 40G MTP breakout cable can provide 96 high-density duplex LC ports for 10G connection, while a 2U 100G MTP breakout patch panel can support up to 160 duplex LC ports.

Reduced Congestion: Reduced cable slack means less clutter, less confusion and an easily organized, better-labeled cabling infrastructure. You can also manage cables in any direction-horizontal or vertical, front or back.

Two Main MTP Breakout Patch Panel Solutions
1U 40G Breakout Patch Panel Supporting Base-8 Connectivity
Base-8 connectivity is supposed to be the most suitable network link which can supports popular 40G switches today and 100/400G networks tomorrow. High density 1U 40G MTP breakout patch panel shown in the following image is designed to connect 40G QSFP+ ports with 8-fiber MTP cables, mapping to the back of the panel, then breaking out as 48x10G on the front with duplex LC fiber cables.

1U 40G Breakout Patch Panel Supporting Base-8 Connectivity
2U 100G MTP Breakout Patch Panel Supporting Ultra High-Density Cabling
100G MTP breakout patch panel as shown below is designed in a standard 2U rack, which has the same working principle as 40G MTP patch panel, but instead of connecting 40GBase-SR4 ports, it connects 100GBase-SR10 ports with 24-fiber MTP cable (10 for Tx, 10 for Rx, leaving the rest 4 fibers unused) to the rear of the panel and then break out as 80x10 on the front with LC fiber cable.

2U 100G MTP Breakout Patch Panel

With more and more high-speed equipment deployed in data center, integrating those different speed network poses a issue to IT managers. MTP breakout patch panel efficiently and conveniently solve this problem. It can support your ability to plan, deploy and upgrade your network to meet the growing demand for additional and higher speed.

FAQs About OM5 Fiber Optic Cable

4. may 2017 at 8:05 | jack
Data centers everywhere are moving quickly to manage ever-increasing bandwidth demands. And the emergence of cloud computing has acted as catalyst for driving even faster adopting of new network technology and higher bandwidth. Speeds as high as 40G and 100G Ethernet have already become mainstream in data centers, and the industry is working collaboratively on next-generation Ethernet development, such as 200G and 400G Ethernet. In this high speed migration, multimode fiber (MMF) plays an important role. As everyone knows, OM1/OM2/OM3/OM4 are commonly used multimode fibers in networking field, especially OM3 and OM4 are proven to be the future-proofing MMF. And now, a new types of MMF fiber medium-OM5, specified in ANSI/TIA-492AAAE and published in June 2016, is introduced. OM5 is being presented as a potential new option for data centers that require greater link distance and higher speeds, however, is it really a good solutions for data centers? This post will deal with this question from some FAQs about OM5.

Q: Does OM5 Offer a Longer Transmission Distance than OM4?
A: Actually, for all current and future multimode IEEE applications including 40GBase-SR4, 100GBase-SR10, 200GBase-SR4, and 400Gbase-SR16, the maximum allowable reach is the same for OM5 as OM4. According to a recently done application testing with 40G-SWDM4 transceivers, it shows that 40G-SWDM4 could reach 400 meters over OM4 cable, while over OM5 cable, the module can achieve link length up to 500 meters. Besides, if a data center is using non-IEEE-compliant 100G-SWDM4 transceivers, it proven that OM5 can support 150-meter reach-only 50 meters more than OM4. In addition, for most data centers, when transmission distance over 100 meters, IT managers will choose single-mode fiber.

transmission distance of OM4 and OM5 in 100G
Q: Does OM5 Costs Less?
A: As the matter of fact, OM5 cabling will costs about 50% more than OM4. Besides, with the considerably declined costs of single-mode transceivers over the past 12-18 month due to silicon photonics technologies and large hyperscale data centers buying in large volumes, more and more users will be pone to choose single-mode transceiver modules. For example, 100GBase-PSM4 using single-mode MTP trunk cable that can support 500-meter reach is only $750.

Q: Is OM5 Really Required for Higher Speeds?
A: All of the IEEE standards in next-generation 100/200/400G Ethernet will work either with SMF and MMF, but in most situations, these next-generation speeds will require single-mode fiber, since IEEE always strives to develop future standards that work with the primary installed base of cabling infrastructure, so customers can easily upgrade to new speeds. Besides, none of these current active IEEE standards addressing next-generation speeds will use SWDM technology.

Q: Will OM5 Create Higher Density from Switch Port?
A: As we all know, it is common in data center using 40GBase-SR4 to increase port density by breaking out 40G to 10G with MTP breakout module or MTP breakout cable. This is also a benefit of new 100GBaes-SR4 modules, which use OM4 cabling. However, if data center manager decides to use 100G SWDM4 modules with OM5 cabling, they cannot breakout into 25Gb/s channels, which will become a real issue as the 25Gb/s ecosystem fully develops and we begin to see more 25G to the server.

According to the questions we have discussed above, it is apparent that OM5 is not suitable for large data centers. As far as I'm concerned, for current high-speed network applications, OM3 and OM4 is still the most recommended multimode fibers.