Just as it did in real life, SDxCentral’s look back at this year’s Nvidia’s GTC event overshadowed our look back at this year’s OFC event, and for good reason as Nvidia’s Bay Area blitz was a major stirrer of all-things digital – including OFC’s Los Angeles happenings – while OFC itself was a major stirrer of all-things reality, and I think we can all agree that today’s reality is a total buzz kill.

Now, let me first state that, yes, like all technology shows nowadays, this year’s OFC event had its share of “number hype.” Terms like 800, 1.6, and 3.2 were thrown around wider and faster than the speed those numbers represent for data flowing through optical network pipes. But those numbers were also tempered by real-world challenges that lacked decimal points.

Most of these challenges were tied to scale-across architectures, which are links that connect geographically diverse compute resources and was a central theme to the OFC event. There was a bit of scale-out involved, but that was mostly theoretical in nature and really more involved with whatever was happening up north.

For instance, one of the coolest new optical fiber technologies coming down the pipe (get it?) is hollow-core fiber (HCF). This is a pretty neat idea that I think came from some smart person that unsuccessfully tried to drink from of a solid plastic rod only to realize that drinking from a straw was a better solution.

As such, a lot of smart people at OFC were hyping HFC as the NBT. (I’m told the kids say this rather than “next big thing.” Oh, wait, I am now being told that kids never said this.)

However, like my attempts at creating cool acronyms, the idea of HCF is proving to be much more compelling than reality.

Jeff Mundt, senior director for optical transport planning at Verizon, during a panel discussion at OFC said that while the carrier is very much interested in HCF, the technology is “350-times” more expensive than current optical fiber and is as fragile as a paper straw. “I literally have a one-kilometer link, and one of our hollow-core fibers is cracked already,” Mundt said.

I’m not that smartest straw in the drawer, but that seems like a bad sign for planned HFC deployments expected to link tens- or hundreds-of-miles between compute resources.

Now, just like how straws have for the most part replaced solid plastic pipes for consuming beverages, I am sure HCF will eventually become the norm, but reality is restraining that replacement cycle.

And should HFC break through that barrier, how will it be deployed? Many pointed at multirail inline amplifiers as the answer, touting the ability to cram more inline amplifiers into deployments to boost performance.

But that cramming comes with logistical challenges that have those tasked with actual deployments on edge and holding their wallets.

“There are some practical questions here of how many lines or rails per rack can I put in, and what's the power and cooling that that's going to be required?” Mundt asked of this option. “And a practical question that we're facing is, if I'm looking at a new fiber build right now, how much land do I buy at all of these ILA (inline amplifier) sites? Am I building and or buying an acre? My real estate team needs to go acquire an acre, because I can plan on putting two shelters and two generators there, or am I going to have to have 10 shelters there for all of these ILAs?”

Interoperability intrigue

Several other OFC panels also touched on the basic topic of interoperability challenges for software and hardware advances. Interoperability is something that an industry as diverse as optical networking and with a topic as vast as scale-across faces on a daily basis.

These OFC interoperability panels were quite insightful, with representatives touching on the benefit of their respective approaches to a solution and how a more focused approach would be preferable. However, these points of view seemed to lose focus when the challenge of what single approach should be taken or even how the industry should approach winnowing down the suspects.

Coming from the old days of telecom standard battles, these conversations were both entertaining and expected. No one wants to give up an inch – or dollar – when it comes to core IP, but at some point someone has to bet a penny in order to potentially win a pound.

This is also a different challenge to what Nvidia and co must face in the scale-out and scale-up world, where Nvidia can throw a few billion dollars at a firm or two to excite some momentum of mass toward their scale-across cause. But, when it comes to scale-across, that investment gets quickly diluted by the size of that mass.

Again, these OFC topics were most amusing after hearing from my colleagues that attended GTC, where every word uttered by Nvidia CEO Jensen Huang – regardless of its real-world-ability – were gospel, fact, and billions of dollars of value creation/destruction. They also directly impacted the mood of OFC attendees, who were gloomy on the first day of the OFC event due to Huang’s lack of scale-across recognition during the first day of GTC, but quickly turned gleeful by the second day when Huang mentioned the scale-across cause during his day two keynote.

I’m telling you, it was weird. It was like having reality controlled by some unseen digital force.

Don’t get me wrong: the OFC event was packed with more than 15,000 super-smart attendees ready to advance optical fiber connectivity in really cool ways. But, real-world challenges – and Nvidia’s shadow – were also front-and-center and are aligning to be significant hurdles toward those hyped advances reaching reality.