Internet 2021 - A Look Forward
What I am keeping my eye on in 2021
Happy New Year! For those who celebrated the new year on January 1. For others, Happy New Year in advance.
What is on my mind as we enter this new year?
Image Source: Pixabay
EU Signs Trade Pact With China
The intent to come to an agreement on investment between Europe and China is something that will be interesting to watch. It is far than clear that it will be ratified by all the EU members, and what the final conditions will be, but nonetheless, the ability of Chinese companies, specifically telecom companies, to do business in the EU has significant implications. Especially given the recent momentum in the direction of blocking companies like Huawei. The consequences may play out for both European and American Networking companies. Perhaps even some non-Chinese Asian networking companies that had the opportunity to capitalize on anti-Chinese sentiment. The fact that European leaders felt energized to push this forward before Biden officially takes office, is perhaps of interest as well.
The Important: What will be the vector of Chinese networking companies in Europe by the end of 2021.
Tech Mission
Science was left with a little bit of a black eye over that whole Atomic bomb thing. The last few years have been a reminder that all technology has downsides as well as upsides. Shssh, don’t tell the luddites.
Arguably, one of the most interesting controversies of 2020 was the medium article penned by Coinbase CEO Brian Armstrong, on being a mission-based company.
“I want Coinbase to be laser focused on achieving its mission, because I believe that this is the way that we can have the biggest impact on the world.”
The perspective Mr. Armstrong expressed, flew in the face of West Coast sentiments on companies needing to be activists on social issues, and also, to some extent, my own musings on the power of the emerging values-based buyer.
As a person who strongly hates politics entering the workplace, or there being any explicit or implicit non mission-based litmus test in the workplace, there was a part of me that immediately resonated with what Mr. Armstrong wrote, especially in his follow-up clarification where he said
“Does Coinbase just stand for making a profit?
No, we stand for accomplishing the mission and for creating a great place to work. Growing revenue and profit is the only way we will be able to grow our team to build all the things that are needed to accomplish the mission. And we need an environment that is welcoming to everyone to attract and retain team member.”
On the other hand, as I said at the beginning of this section, splitting the atom had some consequences, as did the rise of big tech.
The Timnit Gebru incident at Google, which I wrote about here and here, was also a reminder about how shifting societal pressures can easily become a headache for CEOs, resulting in Google’s CEO issuing an apology. Not everyone is willing to call it an apology of course, especially Timnit herself, but I was pleased that Mr. Pichai at least acknowledged how difficult work issues are dealt with must be reviewed. Time will tell how this plays out.
I would still argue that the best way to ensure that minorities are treated with respect, is a commitment to making sure all employees are treated with respect. This is not the same as saying all lives matter. There have clearly been historical barriers for women/minorities, and this has to be dealt with and acknowledged, in a mission-supporting way. However, there was more to the Timnit Gebru incident than just a minority issue, and it is a discussion for all tech workers on how disagreements, especially ethical disagreements, are treated in the workplace.
The Important: The role of Big Tech in society at large is under the microscope, and within tech, the approach of tech companies, to pressing social issues, especially in the self-similar trending West Coast culture, is a cauldron of distraction. And let’s not even get into the much-discussed Texas diaspora. Much for tech CEO’s to balance beyond running a company and making it successful.
Distributed Applications and Personal Privacy
The emergence of various startups focused on distributed applications, definitely has my attention. In addition to the DWEB movement which is an anti-centralization/social dilemma movement as well as an individual empowerment movement. All these trends intersect with the entrepreneurs focused on personal privacy: gaining control back of our own individual information. Many of these trends strike at the heart of big tech business models.
At the same time, Amazon is expanding its footprint by rolling out new availability zones, offering on-prem cloud models, and big tech is responding in a variety of ways to pressures and trends, including Twitter’s recent experimentation with user-controlled private audio chat rooms.
The Important: The mix of evolving big tech models and distributed entrepreneurial efforts is likely to result in compelling innovation over the next few years.
Network AND Security Services
The convergence of network AND security services manifest in the Gartner-penned term SASE (secure access service edge) is probably a bit more of an aspiration than a reality today, especially as the SD-WAN side of the equation is still emerging. However, it is an interesting thesis that has merit.
If security service companies are building world-wide networks with 100+ POPs, why not add in some SD-WAN. If SD-WAN service companies are building world-wide networks with many POPs, why not add in security services. Now, there are also distributed application companies building world-wide networks. Does seem like this segment is ripe for consolidation at some point.
While all this is going on, the lowly network engineer is asking the question “how can I best achieve an enterprise-wide architecture for in-office and WFH employees and partners?” To which the answer might be router and non-router-based SD-WAN solutions, especially if they come with excellent cloud-managed experiences.
The Important: Where will we be on the consolidation of security and networking services as we exit 2021?
Automation & Autonomy, Distributed & Centralized
The Internet is famously based on an ethos of survivability through distribution and network-level redundancy. Yet, the industry finds itself on the precipice of a major shift in mindset.
As we move to architectures that make greater leverage of controllers, it is conceivable that we over-rotate too much in that direction, and for sure, there are some who are concerned that we will do so in a way that disregards the traditional Internet ethos. There may also be some work to do to make sure controllers scale as needed with sufficient availability/survivability - which is all normal technology maturity stuff.
What can be intelligently, and fact fully stated is the following:
Centralized systems may have the most knowledge of mission-based intent and policy.
Centralized systems may have access to information about the network, not currently available in IP routing control planes (queue depth, congestion, CPU/memory utilization, etc.).
All the above can improve network operations and adds up to an era of augmented routing and augmented operations, where it makes sense. I say where it makes sense because at least anecdotally, the feedback I receive is that for small networks, the urgency to move away from current approaches is, like the networks themselves, small. Leadership has come from some of the biggest/highest growth networks in the world, hyperscalers, and their mindset to IT infrastructure is now trickling down.
As that mindset trickles down, it will trickle into different cultures, with varying capacities and capabilities to follow the hyperscaler way.
Just as the centralized only model seemed set in stone, initiatives like RIFT have come along to remind the IP community of its roots: network-based autonomy, where we think of automation and autonomy not at a unit device level, but at a set of devices level (trying very hard not to say “sub net/network”).
The Important: Trends towards network-based autonomy, augmented by centralized analytics, optimizations, and policy, hold the potential to reshape networking away from both the past, and from models that may only work for hyperscale operations with significant operational capacity and capacities. This is a trend that will play out over many years and decades.
Overlays & Underlays
The IP routing world has embarked on a journey to segment routing. The momentum is there. Standards are there, kind of, and Segment Routing has benefits to offer. Whether the Internet is about to take a hard turn towards IPv6, a decades long discussion, is now on the front burner as well. The great Segment Routing compression header debate (SRv6) is doing nothing to clarify this issue. Surely 2021 will be the year where this is put to bed.
BGP as a pervasive overlay technology for services has momentum, makes sense, and I am probably more for it than against it.
At the same time, there is this undercurrent of concern, that focusing the internal implementation architecture of BGP on overlay services may have unintended consequences for how BGP operates as the glue that holds the Internet together, interconnecting networks with different administrative domains.
All this occurs on a foundation of legacy programming languages which are now being supplanted in IT. Mitigated by the inherent survivability of a distributed Internet that is tolerant to individual node failure.
The Important: As overlay and underlay considerations become more complex, how will the industry strive to maintain quality and stability in what is now among the most important economic foundations in the world.
Silicon and Optical Wars
Will 2021 be the year that 400ZR/400ZR+ establishes itself and sets a new direction in access/metro networks?
Will 2021 be the year that Cisco establishes itself as a credible merchant silicon provider?
Will 2021 be the year where Broadcom’s strategy for countering Cisco becomes clear and effective?
Will Nokia’s move to Broadcom in the data center shift into the WAN?
Juniper has also announced new Broadcom-based products for the data center and of course has had a mixed silicon strategy for some time.
Will AI inference engines emerge in network processing silicon, in a more substantial way, in 2021.
The Important: There is much to keep an eye on in the silicon and optical layers during 2021.
Conclusion
Well, that’s probably enough things to be keeping an eye on, though, the list of potentials is longer.
Tech is far from boring. The intersection of culture, business, economics, and how we spend every waking minute of the day, perhaps even our sleeping moments.
Wishing everyone a joyful, healthy, and purposeful new year!
Note: Over the last few weeks my keyboard has become “sticky” repeating many keystrokes. I have made a best-effort to catch them all, but I will need to make a decision in this new year on whether to get a new keyboard/laptop or not.