Overview over 40G Ethernet Interconnection Solutions

The latest trends in the networking world are driving the manufacturers to reach higher speeds and more reliable Internet. Even though 10 Gigabit Ethernet is still on its way to fully take over the world of Datacenters and Service Providers, more and more IT managers are demanding higher speeds and they must consider how they will handle the next generation of Ethernet technologies.  They must consider how to fulfill their client’s needs in aspect of high bandwidth demand, server Virtualization, Cloud computing and backup and their needs for faster Internet and Computing overall of their end customers.

When IEEE officially standardized these 10GB and 40GB solutions they paved the way not only for 40GB Ethernet but for 100GB Ethernet also. The idea of the more powerful 40GB and 100GB Ethernet handling the huge core and aggregation traffic and leaving the 10GB Ethernet to deal with the access layer has slowly but surely started to become a reality.

Many manufacturers will agree that the data transfer rates in the access layers are doubling every 24-30 months, while the data transfer rates in the core and aggregation layers are doubling once every 18-20 months, as research show. This issue, together with the new applications and technologies demanding for high bandwidth speeds are the primary driver leading to 40GB Ethernet.

During the last couple of years many leading IT managers and manufacturers were having second thoughts if they should move to the new technology and solutions that 40GB Ethernet provides or if they should concentrate more on the 100GB and wait for it to become commercially available. However, with 40GB providing design flexibility and cost advantage over the 100GB Ethernet their hesitation might come to an end. Today 40GB Ethernet can safely be deployed in the core and the aggregation layers of Service Providers leaving the 10GB Ethernet for the data traffic in the access layers.

A detailed look at the 40GB Ethernet

Examining the 40GB and 100GB Ethernet and comparing it with each other we can come to a conclusion that they share similar characteristics. Both are Ethernet standards developed by the IEEE 802.ba Task Force so they are able to send frames at 40GB/s and 100GB/s speeds. They also share the same physical layer characteristics when it comes to cabling as they both support copper cabling, Single- mode fibers and Multi- mode fibers. Both standards started developing in June of 2008 and both were officially standardized in June 2010.

 There are some objectives that drove the development of 40GB Ethernet:

• Support of the full duplex operation only
• Preserving the 802.3 Ethernet frame together with the 802.3 media access controller (MAC)
• Providing physical layer support over great distances:
  • over 100 meters on OM3 Multi- mode Fibers and 150 meters on OM4 Multi- mode Fibers
  • over 10 kilometers on Single- mode Fibers
  • at least 7 meters over a copper cable
  • at least 1 meter over a backplane

Because fiber connectivity in higher speeds is being condensed, at the heart of the 40GB Ethernet functioning is a complex system of connected equipment. Most commonly a pair of transceivers connected by Multi- mode or Single- mode OM3 or OM4 active optical cables. These transceivers are simplified as plug and play active equipment interfaces which plug into various devices with the help of interface cards and switches.

40GB Ethernet has high demands regarding the transceivers and the way they are designed and manufactured. These transceivers have been developed following the industry standards:

• CFP or C-Form Factor Pluggable (figure 1) features 12 transmit and 12 receive 10GB/s lanes in order to support one 100GB Ethernet or up to three 40GB Ethernet ports. By its nature it has a larger size so it’s more suitable for Single- mode Fibers, but it can also be used with Multi- mode Fibers or copper cables.
• CXP is a commonly a copper connector which also provides 12 lanes in each direction but due to its smaller size than the CFP it’s used only for Multi- mode optics and copper.
• The QSFP or Quad-Small Form Factor Pluggable (figure 2) is a transceiver similar in size with the CXP. It provides four receive and four transmit lanes and supports 40GB Ethernet. At the moment these transceivers are used mainly for Multi- mode Fibers and copper, however in the future it may support Single- mode fibers also.
• SFP or Small-Form Factor Pluggable transceiver (figure 3) is a compact transceiver used mainly for the 10GB solutions.

When it comes to the cabling for the 40GB Ethernet, cabling can be optical fibers or copper. However the optimal solution for cabling remain the OM3 and OM4 active optical cables (AOC) as shown in figure 4. Both 40GB and 100GB solutions today can be deployed using the same cabling systems. The only difference is between deploying Multi- mode vs Single-mode Fibers depending on the application distance needed.
Multi- mode Fibers will demand the use of parallel optics using Multi-Fiber Push-On (MPO) connectors and this would require additional cable infrastructure to be deployed depending on the system. On the other hand with the Single- mode Fibers the use of serial transmission and use of LC or SC connectors would be a must.

The parallel optics together with MTP interfaces will guarantee that data is transmitted and received simultaneously.In the past the Ethernet standard has used duplex fiber cabling with each channel using one fiber to transmit and one fiber to receive data. However, with the latest 802.ab standard the requirement is to get more traffic with multiple lanes of traffic per channel. In order for this to be done the use of parallel optics is a must. The 40GB Ethernet requires a 12 Fiber cabling solution with each channel having four fibers for transmitting data and four fibers for receiving data. The middle four fibers remain unused.

This approach is based on the demand for higher speeds and these advanced transceivers engineered with the latest technology available would take advantage of the full bandwidth of the laser optics. This would allow the Datacenters and Service Providers to invest in higher speed and higher grade equipment and fiber media. Both Single- mode (SMF) and Multi- mode (MMF) Fibers have been standardized. However, a swap from one to one transceiver used today in 10GB Ethernet solutions with six serial 10GB/s links to six parallel 40GB/s links will require the addition of five new 12 fiber ribbon cables and twelve MTP equipment cords.

When it comes to using OM3 or OM4 Fibers, the OM4 Fibers allow the 40GB Ethernet to reach an additional 60% of the core to aggregation links compared to OM3. The OM4 Multi-mode Fibers would allow additional 30-50 meters of deployment according to the IEEE 802.3ba standard. The initial process for upgrading the existing network infrastructure to 40GB Ethernet solution has already started.

The upgrade has started mainly in the Datacenters and Service Providers, in the layers that connect the access with the aggregation layer.

Considering the gains in productivity and bandwidth, upgrading to 40GB Ethernet would prove to be very cost-effective. The mitigating circumstance when upgrading the network to 40GB Ethernet architecture is the possibility of using the existing 10GB switches and upgrading only their interface line cards and transceivers. Even later in the future when upgrading to 100GB Ethernet would be needed it would be done with deploying a CFP transceiver for extra flexibility during the process.

Nowadays for sure, with the 802.3ba specifications showing the right way, the leading IT managers and manufacturers can safely plan the next step in their chase for higher speeds.