For those looking for some context around Nokia, the service provider for a big WAN I am familiar with (Cox) is all run on Nokia gear. Also, nearly every local police dept I am familiar with runs a Nokia firewall.
If it weren't for geopolitics, Nokia probably would not be around at all at this point.
The world is still visiting the Huawei stop on the journey to consolidated, commodity networking hardware.
Go to your AWS dashboard and set up a VPN gateway. There are about 15 different vendors for which AWS can generate router configuration.
Does any of that make sense? Software has basically consolidated around like 5 platforms for 95% of problems: the browser, Android, AMD64 in data centers running functionally fungible and undifferentiated Linux distributions, Windows on AMD64 on PCs, and iOS native. Gaming consoles & "CUDA" making a strong showing after that, but if that all vanished tomorrow, people could play games on Windows computers and ChatGPT would be way way more expensive.
Then there's a long tail of weirdo crap that will die as quickly as it came to be, supported by people who conflate "existence" with "making sense."
All of networking hardware is weirdo crap platforms! Networking people are going to come out of the woodwork like audiophiles, talking about P4 and eBPF and whatever. I'm sure Nokia Router OS or whatever it's called is very innovative. There used to be a lot of PC vendors too! At the end of the day none of those differentiators are protectible, and they are fundamentally laundering free innovations done by a very small number of people, like Linux Kernel developer. They will not stick around.
> they are fundamentally laundering free innovations done by a very small number of people, like Linux Kernel developer. They will not stick around
There's a big difference between inventing something (Linus Torvalds as the "genius big picture guy" who gets paid in fame) and turning it into a viable business that customers can depend on (the boring "execution guy" who gets paid in money).
It's like how research labs in universities often develop some new technology, and a for-profit firm takes it to market. People get all up in arms about this laundering of free innovations without realizing that the two take very different skill-sets and require very different incentive structures.
My (limited) understanding of heavy network hardware is that it is the way it is because of compatibility.
No one running backbone or DC gear wants to be bitten by edge-case-bug-of-the-month, and so they stick to approved configurations, which are usually vendor-specific for specialized functionality. Because that's what you need to run in order to trigger your "Have a vendor engineer here in 20 minutes" support clause.
There's always a tug between the market (wants vendors to support standards) vs vendors (want market locked into their solutions). As a result, inter-vendor qualification is limited to the minimum they can get away with.
Networking is a combination of hyper-specific hardware solutions "this ASIC targets the mid sized enterprise core and has features specialized to doing that role best, the lower tier version from even the same manufacturer and family however has these limitations... and this carrier focused L2 switch seems similar at a glance but actually has a nearly non-overlapping choice of hardware features baked in" and few (outside of the hyperscaler-like crowd) who actually want to manage that themselves despite knowing no generalized solution is really ready to manage it well either. This results in a massive amount of NOS customization and backwards compatibility that changes horizontally, vertically, and each generation in time. In the meantime, anyone who thinks they can optimize a specific use case better can introduce a new weird device and become popular because it achieves that use case at a great price.
That said, things have been moving in the right direction. E.g. the newer "Nokia Router OS" SR Linux is "just" bog standard Linux with a microservices based service layer which implements industry standard APIs for configuration and telemetry streaming. It still has a lot of legacy stuff, and even a replica of the classic CLI built on top of these APIs, but there is a bit more convergence in that I've been able to deploy Nokia gear and Arista gear in the same DC and manage them as one EVPN fabric with mostly identical API management to get there.
The short version of the above is there are still so many disparate weirdo systems precisely because the use cases of networking are well fit for wildly differentiated hardware every iteration which then drives having specialized software to go with it. Until that advantage goes away I don't see the networking space consolidating all that much.
> At the end of the day none of those differentiators are protectible, and they are fundamentally laundering free innovations done by a very small number of people, like Linux Kernel developer.
What does this use of laundering mean? The financial application of laundering does not make sense to me here (to make something procured illegally look like it was procured legally).
Networking equipment used by ISPs (which is Nokia's main revenue source) is not really comparable to consumer level stuff.
That said I left telecom (used to work for Huawei btw) in 2018, people were talking about hyper convergence and running network functions in K8s and shit. Not sure how much of that materialized.
I wonder what would drive that market down. Cell network equipment is an infrastructure investment, right? It seems they are predicting a pretty long-term demand drop in that case…
It may surprise some to know that many network operators buy backbone hardware platforms (DWDM transmission, cell carrier systems, etc) configure, operate, field-service, and even finance these lumps of custom hardware with their ASICS and line cards (plus standard CPUs and regular network interfaces) via the equipment vendor. They buy them with money loaned from the vendor (lower cost of capital), design and config the network with vendor people, and keep them running with people wearing Nokia, Huawei, Ericsson socks. These systems tend to have quite a long lifetime in the field.
I remember reading that the move to 5G would make cellular equipment more like software with upgrades and needing less hardware replacements to support newer tech.
Nokia is one of the top cellular/mobile backend hardware providers. 5G is turning out to be a lot more expensive than projected, the rollout has slowed, so equipment makers like Nokia are feeling the pain.
Readit News
The world is still visiting the Huawei stop on the journey to consolidated, commodity networking hardware.
Go to your AWS dashboard and set up a VPN gateway. There are about 15 different vendors for which AWS can generate router configuration.
Does any of that make sense? Software has basically consolidated around like 5 platforms for 95% of problems: the browser, Android, AMD64 in data centers running functionally fungible and undifferentiated Linux distributions, Windows on AMD64 on PCs, and iOS native. Gaming consoles & "CUDA" making a strong showing after that, but if that all vanished tomorrow, people could play games on Windows computers and ChatGPT would be way way more expensive.
Then there's a long tail of weirdo crap that will die as quickly as it came to be, supported by people who conflate "existence" with "making sense."
All of networking hardware is weirdo crap platforms! Networking people are going to come out of the woodwork like audiophiles, talking about P4 and eBPF and whatever. I'm sure Nokia Router OS or whatever it's called is very innovative. There used to be a lot of PC vendors too! At the end of the day none of those differentiators are protectible, and they are fundamentally laundering free innovations done by a very small number of people, like Linux Kernel developer. They will not stick around.
There's a big difference between inventing something (Linus Torvalds as the "genius big picture guy" who gets paid in fame) and turning it into a viable business that customers can depend on (the boring "execution guy" who gets paid in money).
It's like how research labs in universities often develop some new technology, and a for-profit firm takes it to market. People get all up in arms about this laundering of free innovations without realizing that the two take very different skill-sets and require very different incentive structures.
No one running backbone or DC gear wants to be bitten by edge-case-bug-of-the-month, and so they stick to approved configurations, which are usually vendor-specific for specialized functionality. Because that's what you need to run in order to trigger your "Have a vendor engineer here in 20 minutes" support clause.
There's always a tug between the market (wants vendors to support standards) vs vendors (want market locked into their solutions). As a result, inter-vendor qualification is limited to the minimum they can get away with.
Someone tell me if I'm wrong, though!
That said, things have been moving in the right direction. E.g. the newer "Nokia Router OS" SR Linux is "just" bog standard Linux with a microservices based service layer which implements industry standard APIs for configuration and telemetry streaming. It still has a lot of legacy stuff, and even a replica of the classic CLI built on top of these APIs, but there is a bit more convergence in that I've been able to deploy Nokia gear and Arista gear in the same DC and manage them as one EVPN fabric with mostly identical API management to get there.
The short version of the above is there are still so many disparate weirdo systems precisely because the use cases of networking are well fit for wildly differentiated hardware every iteration which then drives having specialized software to go with it. Until that advantage goes away I don't see the networking space consolidating all that much.
What does this use of laundering mean? The financial application of laundering does not make sense to me here (to make something procured illegally look like it was procured legally).
That said I left telecom (used to work for Huawei btw) in 2018, people were talking about hyper convergence and running network functions in K8s and shit. Not sure how much of that materialized.
Nokia makes cellular network equipment. HMD Global designs and markets smartphones with the Nokia brand.
Deleted Comment
Lots of discussion last week: https://news.ycombinator.com/item?id=37940317
US demand shrinks? Growth uncertain? Did nokia carve out a niche in the US market because I haven't seen a nokia phone or any nokia product in ages.
Nokia makes cellular network equipment. HMD Global designs and markets smartphones with the Nokia brand.
:)