Hacker News Reader

Comment by femto | Hacker News Reader

> ...as the company transition from its hardware roots into a software-centric organization.

With all these companies exiting the hardware space, who is going to be doing the hardware?

Will hardware remain a niche department within these software based companies? Are off-the-shelf systems-on-a-chip now powerful enough that there is now no need for hardware, beyond a generic universal SoC based device? Is it going to be an unserviced niche for new players to move into?

To use the specific example of Cisco: what hardware are Cisco planning to have their software running on in (for example) 5 years time?

View on HN

The software will run on inexpensive dumb hardware from whitelabel manufacturers such as Quanta computers. The secret sauce in hardware that gave Cisco big margins is increasingly replaced by software that is good enough (e.g. You can use SDN controller and a cheap switch hardware to replace very expensive Cisco stuff). The trend in Enterprises and telcos is to use cloud (AWS, GCE, Azure,...) and open source (aka Facebook Open Compute) so the hardware spending has slowed down. (in tier 1 carrier I deal with, all new deployments must be on Openstack that uses whitelabel hardware purchased in bulk). EMC, Oracle, Cisco, HP are facing the impacts of such decisions.

> all new deployments must be on Openstack

you have my sympathies

Better that then yet more money funneling into the pockets of the big cloud companies.

How so? Most of the OpenStack stewards are as stodgy and corporate...

OpenStack is open source. It can be forked, modified, whatever. Let me know when you manage to build your own AWS cloud with Amazon's software and then I'll entertain the thought that they are comparable.

Its not about money funneling into the pockets of the big cloud providers, its about engineering hours being better spent. Openstack is bit of a mess.

Absolutely. I'm still waiting for any two things in OpenStack to be consistent with each other.

Related: My compilation of OpenStack rage tweets. https://twitter.com/search?q=@stefanmajewsky%20#openstack&sr...

Link just takes me to your twitter profile. I suspect if you can't correctly paste a link, you can't use openstack. :P

Gah, Twitter. Correct link: https://twitter.com/search?q=@stefanmajewsky%20%23openstack&...

What's so bad about OpenStack?

What's good about it? It's a bunch of different dumpster fires created by a bunch of committees across a bunch of companies which mostly compete with each other, and as such have no real incentive or motivation to have it all work well together even if they had the ability to, and the ability alone is seriously in question.

The Vision: a community-managed set of integrated tools that let you create a featureful cloud on hardware you control.

The Reality: a bunch of python scripts everyone forks.. committee politics...frustration

OpenStack is a great example of going "community" before the problem is well understood or there is a solution people trust.

It's basically a tire fire at this point and most of the community participants are market losers.

Substitute Python for bash and I get the same sense from Kubernetes.. I'm not sure it's fair to say "before the problem is well understood... But maybe. Whatever Google had going on internally Kubernetes is very much figuring it out along the way at almost every level. If you are wanting to use on AWS then you have kops, kube-up, hyperkube, protokube; a dizzying, un-approachable and interconnected ecosystem of launch scripts/programs. Very strange interop decisions with AWS, but hard to fault because it's darn near one person trying to manage all AWS integrations.

bash? Isn't Kubernetes predominantly Golang based?

"Substitute Python for bash and I get the same sense from Kubernetes"

Huh? Kubernetes is written in Golang.

the annoying and stupid modularity of it ...

What alternatives are there?

AWS, Azure or GCE....or if you insist on your own hw, just what spit and bailing wire you minimally require... at least you will own and understand it.

Those are not alternatives. They require you to run your software on someone else's computer.

I tried to use OpenStack 4 years in our new PaaS product. I ended up pulling it and using the regular, well baked tools built into our managed hardware. I see much hasn't changed in those intervening years.

>I see much hasn't changed in those intervening years.

Source?

White box switches are all great in the data center if we are talking about a shitload of layer 2 and TRILL, but somebody still has to manufacture the core routers capable of advanced IP layer 3 features at line rate 400-1000Gbps bus backplane per slot. You're not going to whitebox replace an ASR9010 with 3rd generation line cards.

It's a much smaller market than the old "everything with more than one network port should be Cisco" mantra, though. Yeah, Cisco won't disappear any time soon, but their market is shrinking.

> Yeah, Cisco won't disappear any time soon, but their market is shrinking.

do you think cisco is in danger of becoming SGI of networking ?

A much smaller company could definitely create these very high end switches and routers, with a business and support model unlike that of Cisco.

And they won't be trusted or used by large carriers for serious POPs for many years. New manufacturers have tried to show up on the market and sell routers that are designed to be highly redundant and offer a full suite of layer 3 feature set to handle full global v6/v6 routing tables. They've either failed or been acquired.

Major ISP backbones are very risk averse, when you're trying to run a six nines availability IP services POP. The routing software needs to be rock solid and mature when you're looking at a twin pair of core routers in a node that might be handling 200-300 Gbps of traffic and critical edge peering/transit connections.

> They've either failed or been acquired.

That is certainly not the end of such endeavours. Cisco can't buy ALL of them forever.

Many would argue that M&A is Cisco's primary market.

My counterargument is that Tomahawk LC (the third generation LCs referenced by parent) are just NP5 and Mellanox just picked up EzChip. I imagine you'll see whitebox competitors targeting large swaths that use ASR9K for the datapath resources it provides (perhaps Mellanox will even build this box).

Others, such as in my particular space, were previously forced to ASR9K or similar for scale reasons (because scaling full tables is hard when you need capacity too), but Broadcom's Jericho chips hit the scale mark with added TCAM, kicking us from $200K-ish/PoP for barely meeting needs to $60K/PoP way over provisioned in the case of NCS5501 or Arista 7280R. Of course, there are caveats to how this works out, but good enough is good enough!

Customer mix is important for margins. SPs, for example, are notorious for getting bottom dollar deals for their scale and ability to play the other side. And sure, they need the feature set of ASR9K. But when the rest of the mix (higher margin) looks around, they see that the scale promises are unnecessary for their operations and move on to lower cost boxes.

The networking field, as with the server hardware field, is seeing a movement towards commodity hardware. Custom network silicon is expensive to develop, and no longer is returning the investment that it did. This is cause companies like Juniper and Cisco to look at ways to expand their profits, while essentially making nothing on the hardware.

Cisco along with others are looking by at things like licensing the IOS to run on any device. So a company could buy 50 white box switches and slap IOS or JUNOS on them.

In the networking world OSs are even more religious then the Microsoft Linux debates of the past. I know networking shops that will never buy anything but Cisco because they only know how to network on Cisco and they refuse to learn anything else. That is the market that Cisco is going to target.

> Cisco along with others are looking by at things like licensing the IOS to run on any device.

ios is a custom-rtos for a custom machine (running on mips mostly). it runs control plane for your routers/switches etc. why would you want to run it on a non-csco/vanilla x86 boxes ? unless you can _somehow_ make control-plane h/w agnostic (easier said than done) i don't really see much value in doing this.

You may not. But large enterprises that pay billions a year for networking gear, who can save a billion by buying whitebox, and still get that Cisco feel that they have spent years training and acquiring expertise will and already are. It may not be IOS. But Cisco will come out with something that feels and operates like IOS that runs on x86. Juniper has already ported it's JUNOS to run on x86.

Between whitebox and SDN. These companies won't be selling hardware for the 1-200% markup the have enjoyed all these years.

> Juniper has already ported it's JUNOS to run on x86.

huh :) i was/am under the impression that with junos, juniper started with freebsd as the core, and then added lot's of stuff to it, specifically in control-plane and a bunch of kernel extensions as well.

if that is true, they were already x86 compatible...

with cisco things are very different. i don't think ios ever ran on a bare-metal x86. this is not for the lack of trying though (if you know folks who have worked on ios, check with them, writing about it here, doesn't seem right)

Junos has run on x86 for years - Juniper had i686 control planes in some devices. As well as PPC and a few different kinds of MIPS (Octeon, XLR) - although those are usually data plane.

Newer products use virtualization, i.e. running the FreeBSD-based Junos control plane in a VM.

The very first juniper ever was basically a freebsd box glued to a bunch of custom ASIC and line cards. The m40. It has always been x86 and now x86-64.

Not much difference I'm afraid. IOS-XE is essentially the product of porting the classic IOS and runs it as a controlling process that sits top of linux, and with dedicated forwarding path.

If you look up for IOU, which stands for IOs over Unix, it has been around for quit a long time.

I've seen some older Cisco storage switches which utilised Pentium 4 CPUs and SAN-OS instead of the usual custom built Motorola processors for the non-ASIC components of the device.

> ios is a custom-rtos for a custom machine (running on mips mostly).

this is true for legacy IOS, the "new" IOS-XE and IOS-XR could run on top of linux, in fact IOS-XR 6.0 could run in a linux container: http://www.cisco.com/assets/global/DK/seminarer/pdfs/XR60.pd...

>. why would you want to run it on a non-csco/vanilla x86 boxes ?

AWS, for instance, with the CRS1000V. Though, I think that is IOS-XE and Linux based. IIRC, they are moving to IOS-XE/XR which is more Linux based and can be run on COTS boxes.

|> With all these companies exiting the hardware space, who is going to be doing the hardware?

The manufacturers that have been actually making the hardware for the last 15 or more years, that's who. The only difference in many cases is the color and brand on the case.

I expect eventually margins on hardware will come up and new players will enter. Maybe not, but the siren song of software's high margins and flexibility is only going to last so long. Innovation in hardware is going to slow down a lot, and software people are going to have to start worrying about efficiency again, and will no longer be as able to shrug optimization off onto Moore's Law or abstract away the hardware altogether. Some people will come in with new hardware, offering to alleviate those pains, and we'll have a healthy hardware market again.

Or I'm totally wrong :)

Edge Core builds something like 70% of all big names switches. You can buy an Edge-Core branded switch for a fraction of the price of a Cisco one, and run various OSes on it, mostly Linux-based (Pica8, Cumulus...). You can also run nmap, tcpdump, or whatever you want on it... Once you've tried it you won't look back :)

> With all these companies exiting the hardware space, who is going to be doing the hardware?

There's a ton of smaller players ready to jump in. They're not household names, but they have the expertise to fill the gap. I wouldn't worry about it. Ubiquiti, Mikrotik, etc

Ubiquiti is the shit. Really clean APs.

> They're not household names, but they have the expertise to fill the gap. I wouldn't worry about it. Ubiquiti, Mikrotik, etc

These companies have nothing to do with the businesses that make Cisco its real money.

Growth in compute is from the cloud companies. They can purchase switches from ODMs like Edge-Core, Quanta, and Celestica directly. At the end of the day, because switching has largely converged on the same merchant silicon (Broadcom Trident/Tomahawk now, perhaps Barefoot Tofino in the future), they end up with a similar product at a lower price point. Either these companies are big enough that they have their own switch operating systems or they license one.

It's simply a different world than a few years ago.

Much of their hardware is just commodity x86 or ARM parts.

hardware way cheaper to do in china

So is software. What does Cisco actually expect its business to be in 5 years or 10 years? It's the next Yahoo.

Not really cheap for software... I worked for Cisco in Shanghai and I recall we were more expensive than our Ottawa friends.

Reply on news.ycombinator.com