100G QSFP28 Optical Transceivers: Am I using the right tool for the job?

Probably the most common question we get from our customers when looking at our vast array of 100G QSFP28 Optical Transceivers to help optimize their network, is:

“Why are there so many different types?”

One might ask the same question (I know I have!) of the vast array of screwdrivers in a pro mechanic’s toolbox. Just as different screwdriver types (slotted, Phillips, TORX, and so on) are engineered to transfer energy more effectively from a screwdriver to the screw based on a given application,100G QSFP28 Optical Transceiver types are created with unique features to address the needs of specific network applications.

Knowing the tools – and more importantly when and where to use them – will save you valuable time, effort, and money. Here are some of the fundamental tools of the trade every network engineer should be familiar with.

QSFP28 SR4 and LR4

For most network engineers, the 100G SR4 and LR4 transceivers are those tools they reach for most often. The short reach SR4 is the 100G workhorse within head-ends and datacenters, using cost-effective 850nm lasers over parallel multimode fiber (MMF) for distances up to 70m (OM3), 100m (OM4), or 150m (OM5).

The “long reach” LR4s are the other 100G workhorse for applications up to 10km, using lasers in the 1310nm range over duplex single mode fiber. These two definitely make up the volume of 100G transceiver deployments, but there are other tools in the QSFP28 “kit”.

QSFP28 SWDM4

The 100G SWDM4 transceivers are based on short wavelength division multiplexing. Instead of the four parallel lanes of 850nm used for SR4, SWDM4 multiplexes four separate wavelengths (850, 880, 910, and 940nm) to be then transported up to 150m (OM5) over a duplex MMF pair.

The SWDM4 “tool” gives a network engineer the option of upgrading their 10GBASE-SR duplex links to 100G without the need to run new parallel fiber, saving cost and time by leveraging off the existing cable plant.

QSFP28 CWDM4

The 100G CWDM4 provides a lower-cost alternative to the LR4 for links up to 2km. That rather large gap in the portfolio between 150m (SR4) and 10km (LR4) is well served by the 2km CWDM4.

As the name indicates, the 100G CWDM4 uses lasers based on the CWDM grid (1270, 1290, 1310, and 1330nm) as opposed to the LR4, which is based on the LAN-WDM spacing (1295, 1300, 1304, and 1309nm). Relaxing the requirements on the CWDM4 specs allows for a lower cost 100G solution optimized for 2km spans over duplex SMF.

QSFP28 PSM4

The 100G PSM4 provides the 10km singlemode reach of an LR optic, but with an MPO connector to provide a four-lane Parallel Single Mode (PSM) interface, rather than the duplex LC connector on the QSFP28 LR4. A network engineer thus has the ability to break out from a single QSFP28 100G port to 4xSFP28 25G ports, up to 10km away. The PSM4 module helps maximize faceplate density on your core and access routers when aggregating 25G links from subtending equipment. Very useful as wireless and FTTx starts migrating from 10G to 25G ports.

QSFP28 ER4/ER4L

The initial 100G 40km transceiver solutions were based on CFP/CFP2 ER4, and the higher module power capacity of the CFP form factors allowed for the higher power laser and APD (receiver) required for 40G links.

The initial QSFP28-based 40km solution was the ER4L (ER4 “lite”), which relaxed the optical power budget specifications in order to meet the QSFP28 module power consumption limitation. The “lite” meant that to reach 40km, the host switch/router would need to apply FEC (forward error correction) to meet the link requirements.

Thanks to advancements in component integration and die shrinks, a true 100G ER4 is now also available in the QSFP28 form-factor. What does this mean, and why would we want this when we already have the ER4L?

Latency is the primary drawback of FEC, as that additional bit of error correction coding does take time. For latency sensitive applications, having a true ER4 can be an essential tool in your kit.

QSFP28 ZR4

The 100G ZR4 was provides that eagerly awaited upgrade path from 10GBASE-ZR to 100G. Getting four lanes of 25G to reach 80km was no trivial task, and it required the addition of a silicon optical amplifier (SOA) as well as dispersion compensation within the module to meet the 80km link budget.

The resulting power consumption of the 100G ZR4 is thus higher (~5-6W) than its shorter reach counterparts, but it is an invaluable tool in upgrading those long 10G spans to 100G without the need for additional hops or external amplifiers.

QSFP28 DR1/FR1/LR1

Single-lambda QSFP28s are a perfect complement to the 400G QSFP-DDs for breakout applications, going from 400G to 4x100G. The QSFP28 DR1 (500m), FR1 (2km) and LR1 (10km) use a single laser with PAM4 modulation at 50Gbps which allows them to link with their 400G counterparts on a lane-by-lane basis.

Summary  

Going back to the question, “Why are there so many different types?” each type of 100G transceiver is a tool that can help a network engineer optimize their optical network topology for cost, latency, and efficiency. Sure, you might be able to get away with a single screwdriver for a DIY project, but like the professionals, your prized optical network deserves the right tools for the job.

And based on the customer testimonials we’ve received, I don’t think any of these successful businesses would recommend you take on those major risks either. If you’re interested in additional support or want to discuss ways to optimize your network with the right tools and approach, reach out to your Integra sales or engineering team.