cisco transceiver modules
The popular QSFP+ 40Gb bi-directional (BiDi) transceiver enabled data center operators to upgrade from 10Gb to 40Gb without the need to replace fiber cable infrastructure.
The next generation dual-rate BiDi, which has both 40Gb and 100Gb operational modes, offers the same benefits and more.
The Cisco QSFP 40/100 Gb dual-rate BiDi transceiver can function as a 40Gb BiDi, matching its reach of 100m and 150m over OM3 and OM4 multi-mode fiber, respectively.
In 100Gb mode, the dual-rate BiDi’s reach is 70m and 100m on OM3 and OM4, respectively. In both modes, the BiDi reach matches its SR4 counterpart.
As a single transceiver with these two modes of operation, the Cisco dual-rate BiDi enables data center operators to re-use their existing duplex LC-connectorized MMF infrastructure for migration from 10 Gb to either 40 Gb or 100 Gb Ethernet connectivity, or migration from 40Gb to 100Gb. Figure2 shows the 100G BiDi Transceivers deployed in a leaf-spine logical architecture.
• 40Gb and 100Gb dual-rate capability
• Backward compatibility with 40Gb QSFP+ BiDi
• QSFP28 form factor
• Operates over duplex LC-connectorized MMF fiber infrastructure, same as 10Gb SR and 40Gb BiDi
• Onboard forward error correction
• PAM4 optical modulation
• Bi-directional optical technology
• Enables incremental upgrade to 100Gb from 10Gb SR or 40Gb BiDi
Backward compatible with Cisco’s 40Gb BiDi technology
Only Cisco offers backward compatibility with native 40Gb BiDi transceivers in a QSFP28 form factor. Like the native 40Gb BiDi, the dual-rate 40/100 Gb BiDi transmits bi-directionally on both fibers, using nominal wavelengths of 850nm and 910nm and passive wavelength division multiplexing (WDM) filters at either end to isolate Tx and Rx channels. In 40Gb mode, it transmits and receives 20Gb NRZ channels on each fiber, for a total aggregate bandwidth of 40Gb.
In 100Gb mode, it operates 50Gb PAM4 channels, for a total aggregate bandwidth of 100Gb. PAM4 technology enables 50Gb data rate with signaling at 25Gbaud rates. The 40/100G BiDi contains a gearbox to translate the signal from a 4x25G format, native to the QSFP28 form factor, to the 2x50Gb format for the optical domain. It also employs onboard forward-error-correction (FEC) to reduce bit error rate. This makes it possible to use current generation transmitter and receiver technology.
The dual-rate 40/100 Gb BiDi offers multiple migration options to Cisco customers who are currently operating either 10Gb or 40Gb links in their data center architectures. By upgrading network equipment one end at a time, operators can exercise flexibility in schedules and budgets. Figure 3 shows combinations of port and BiDi transceiver types.
Figure1. 40/100G QSFP BiDi transceiver module
Figure2. Example of 40/100G QSFP BiDi in a leaf-spine architecture with Cisco Nexus 9k switches
Figure3. Interoperability with 40G BiDi enables incremental upgrade of network gear.
25G Speeds Up Data Centers and Campus Backbones NOW. With the massive increase in demand for data, equipment providers are responding with 25Gbps edge devices that require more bandwidth than can be provided on a traditional 10Gbps interface.
Whether it’s a server or a campus backbone, high speed data needs to be delivered cost-effectively in a small and low-power package.
In these bandwidth-intensive applications, the choice to go with 25G is clear. To get the same or better bandwidth, the number of 10G interfaces must be 3x (6x for redundancy) or the application needs to move to the larger, more expensive and power-hungry 40G QSFP.
SFP28: For 25G the dominant form factor is SFP28. The SFP28 standard relies on the 10G SFP+ (Small Form Factor Pluggable) standard for mechanical specifications, and the electrical specifications have been improved from one 10Gbps lane that operates at 10.312Gbps to one 28Gbps lane that operates at 25Gbps + error correction. 25G transceivers can be plugged into SFP+ sockets and 10G transceivers can be plugged into SFP28 sockets because they have the same electrical and mechanical pin-out, however the associated host needs to have the software support for associated devices.
Cisco’s 25G transceiver choices include 25G Copper DAC (Direct Attached Cables), 25G AOC (Active Optical Cables) and 25G SR-S (Short Reach) transceivers.
These 25G devices are plugged into Cisco’s data center, campus and service provider switches and routers to provide high speed 25Gbps connectivity. See Cisco’s 25G compatibility matrix for currently supported devices .
25G DACs are generally used in data center applications and provide the lowest cost fixed length interconnect for TOR (Top of Rack) switches to high-performance servers. Depending upon the bandwidth and distance, DACs can be either passive or active and are generally based on Twin-AX cable. For 25G, DACs can generally operate up to 5 meters without active components in the data path. Up to 2 meters, no FEC (Forward Error Correction) is needed. For 3 meters FC-FEC (Fire Code Forward Error Correction) is needed, and for 5 meters RS-FEC (Reed Solomon Forward Error Correction) is needed to correct errors. Generally, at 25Gbps beyond 5 meters, active components are needed in the data path to amplify and correct the signal. These components drive up cost which causes network designers to consider optical interfaces.
25G AOCs also provide a cost effect solution for those same data center applications that require longer distances than 5m. Generally, AOCs are provided in standard lengths of 1m, 2m, 3m, 5m and 10m. However, they are usually limited to about 25 meters because of inventory stock and slack storage issues. Often a data center will be wired with only AOCs for consistency reasons, instead of a combination of AOCs and DACs.
25G-SR is used with standard OM3 or OM4 multimode fiber and is suitable for:
• Data centers that require up 100 meters over OM4 fiber or 70 meters over OM3 fiber for interconnect between TOR switches and leaf or spine switches.
• Breakout configurations in conjunction with 100G-SR4 transceivers where the distances are less than 100 meters for OM4 fiber or 70 meters for OM3 fiber.
• Campus backbones, where the distances between distribution and aggregation switches are less than 100 meters for OM4 fiber or 70 meters for OM3 fiber.
Learn more about how Cisco’s 25G transceiver products are transforming the industry here
Original article from https://blogs.cisco.com/sp/too-slow-25g-speeds-up-data-centers-and-campus-backbones