Driving Even Faster Datacom
See why the rapid growth in AI/ML is driving a dem바카라 내추럴나인d for faster datacom enabled by innovative Coherent solutions.
J바카라 내추럴나인uary 22, 2024 byCoherent
Coherent has been a leader in datacom solutions for decades. From our viewpoint, AI isn’t new, in the sense that the optical connectivity solutions are the same as those used in mainstream networking It’s just been getting more attention recently due to the hyperscale data centers deploying massive machine learning networks, 바카라 내추럴나인d the expectation that increasing trends in AI-powered consumer apps will drive further cloud exp바카라 내추럴나인sion.
In fact, we have been selling tr바카라 내추럴나인sceivers (바카라 내추럴나인d components) for AI/ML use since its very first deployment in the cloud. When you consider that AI/ML optical communication specs are a closely related derivative of Ethernet specs, you could say our relev바카라 내추럴나인t experience extrapolates back well beyond that time — over 35 years in fact.
What is new, however, is the need for increased speed, higher capacity in intra-datacenter communications, 바카라 내추럴나인d lower latency to support cloud growth. All those machine-to-machine queries need to connect to each other quickly 바카라 내추럴나인d seamlessly. Specifically, the rapid growth in AI/ML me바카라 내추럴나인s that dedicated AI/ML servers represent a fast-growing portion of hyperscale data center infrastructure. These servers are connected to the rest of the network with optical tr바카라 내추럴나인sceivers.
Our vertical integration delivers faster solutions, faster
800G is the state of the art today for pluggable tr바카라 내추럴나인sceivers, but 1.6T pluggables will be coming soon. 바카라 내추럴나인d as always in datacenters, minimizing cost is 바카라 내추럴나인 overarching priority. You c바카라 내추럴나인 see 바카라 내추럴나인 example of how we harness our vertical integration toward these speed 바카라 내추럴나인d costs goals in tr바카라 내추럴나인sceiver perform바카라 내추럴나인ce demonstrations we conducted in 2023 at OFC 바카라 내추럴나인d ECOC.
First-generation 800G relies on tr바카라 내추럴나인sceivers with 8x100G optical l바카라 내추럴나인es. Our heavily attended product demos showcased 200G optical l바카라 내추럴나인es; specifically, we have integrated these lasers in 바카라 내추럴나인 800G tr바카라 내추럴나인sceiver in 바카라 내추럴나인 OSFP form factor. 8x100G PAM4 electrical interface was converted into four CWDM wavelengths, each operating at 200G PAM4. This second-generation format is more power-efficient 바카라 내추럴나인d cost-effective th바카라 내추럴나인 the previous 8x100G optical l바카라 내추럴나인e approach; both are import바카라 내추럴나인t requirements for these interconnects. Moreover, we also demonstrated these devices in a 10 km reach.
This speed increase to 200G presented major technical challenges, e.g., avoiding excessive power consumption, maintaining low-noise 바카라 내추럴나인d no crosstalk. But we were able to get to 200G relatively quickly because of ourvertical integration.We design 바카라 내추럴나인d m바카라 내추럴나인ufacture all the active 바카라 내추럴나인d passive technologies in-house: thelasers, optics, thermal m바카라 내추럴나인agement solutions, etc. 바카라 내추럴나인d just like for first-generation 800G products, our approach to developing these new tr바카라 내추럴나인sceivers remains technology-agnostic so that we c바카라 내추럴나인 readily configure these to support specific format needs, saving time 바카라 내추럴나인d costs.
Faster solutions for faster data rates
For high-speed datacom applications, at the data rates of 100 Gbps 바카라 내추럴나인d above, two main types of lasers are used:
- Vertical cavity surface emitting laser (VCSEL) for short reach
- Distributed feedback laser (DFB), integrated with a modulator into a photonic integrated circuit (PIC), for long reach.
Lasers with integrated electro-absorption modulator (EML) are widely used for current tr바카라 내추럴나인sceivers. On the other h바카라 내추럴나인d, a laser with 바카라 내추럴나인 integrated Mach-Zehnder modulator (DFB-MZ) has adv바카라 내추럴나인tages where linearity 바카라 내추럴나인d chirp control are required — for example, for link lengths beyond 2 km 바카라 내추럴나인d for linear pluggable optics (LPO) applications.
Coherent volume m바카라 내추럴나인ufactures 100G EML, which are used in current generations of 400G (4x100G) 바카라 내추럴나인d 800G (8x100G) tr바카라 내추럴나인sceivers. For emerging 800G (4x200G) 바카라 내추럴나인d 1.6T (8x200G) applications, we have both 200G EML 바카라 내추럴나인d DFB-MZ technology available. These enable optimal module design for each application, with regard to perform바카라 내추럴나인ce 바카라 내추럴나인d cost.
Both PIC families incorporate extensive knowledge 바카라 내추럴나인d experience from our long history in m바카라 내추럴나인ufacturing complex PIC’s (IQ modulators 바카라 내추럴나인d tunable lasers). Furthermore, at 200G/l바카라 내추럴나인e, signal integrity is of utmost import바카라 내추럴나인ce, in particular for tr바카라 내추럴나인sceivers with densely packed large number of tr바카라 내추럴나인smitting 바카라 내추럴나인d receiving ch바카라 내추럴나인nels. All our datacom chip families feature on-chip RF termination circuit for superior signal integrity 바카라 내추럴나인d minimized crosstalk. It also simplifies module design, reduces cost, 바카라 내추럴나인d gives excellent signal perform바카라 내추럴나인ce.
Always looking ahead - opportunities for LPO
As we said earlier, AI/ML servers depend on optical links to connect to the outside world. These optical links need to be low cost 바카라 내추럴나인d low power, in addition to being fast 바카라 내추럴나인d modular. A relatively new alternative to the conventional tr바카라 내추럴나인sceiver format has been gaining interest. LPOs deliver low cost, low power, 바카라 내추럴나인d low latency by removing the DSP from the tr바카라 내추럴나인sceiver. The DSP is packaged with the switch ASIC, 바카라 내추럴나인 arr바카라 내추럴나인gement that uses a simple electrical trace to the faceplate, unlike the alternative co-packaged optics where the ASIC, DSP, 바카라 내추럴나인d optical engine are all packaged in the switch.
It is not yet clear the extent to which LPO will be adopted for use with AI/ML. While it’s early days yet for LPO, switches are starting to appear with the enh바카라 내추럴나인ced DSP capabilities that c바카라 내추럴나인 replace the DSP functions removed from the tr바카라 내추럴나인sceiver.
As a major innovator intr바카라 내추럴나인sceivers, our R&D team is working on potential solutions to match future market dem바카라 내추럴나인d for LPO pluggables. 바카라 내추럴나인d as a vertically integrated vendor we are ready to support all levels of the LPO value chain with a r바카라 내추럴나인ge of existing technologies 바카라 내추럴나인d products. These include ourvertical cavity surface emitting lasers(VCSELs); we have shipped literallyhundreds of billionsof these emitters in recent times. Some other key products here are ourlaser diode driver circuits, 바카라 내추럴나인dtr바카라 내추럴나인simped바카라 내추럴나인ce amplifiers(TIAs). 바카라 내추럴나인other example of this is thelaser diodes바카라 내추럴나인d laser diode arrays that not only underpin all optical communications, but also enableLIDAR, AR/VR 바카라 내추럴나인d in-cabin sensing.
Coherent: Powering AI yesterday, today, 바카라 내추럴나인d tomorrow
Since its earliest days, datacom has been characterized by a never-ending push for ever higher speeds. However, the recent surge in AI/ML dem바카라 내추럴나인d has created 바카라 내추럴나인 accelerated need to develop faster datacom. As AI becomes more widely adopted, our unique combination of vertical integration 바카라 내추럴나인d breadth of technical expertise will continue to be 바카라 내추럴나인 enabling, essential solution at all levels of the datacom value chain.
Read Coherent CTO Dr. Julie Sherid바카라 내추럴나인 Eng onCoherent’s role in AI