AC3200 Routers By July?

[thiggins]

if you look at where Linksys/Asus are right now with their current high-end offerings, they're not making a lot of money on them...

Actually, I think they are. They find they can sell fewer units and make as much, if not more than selling lower-cost routers.

 

[thiggins]

That sucks, still no 10 gigabit ethernet.1 gigabit is really far too old and obsolete. (we are pretty late into 2014 now)

There is no business case for 10GbE for consumers. Retail price of a single port copper 10GbE NIC is $300 - $400. Let's say material cost would be $100.

How many people would pay $400 for a router? Linksys already considered the $300 price point for the WRT1900AC and backed off before they started to sell it.

And don't forget additional thermal load.

BTW, please make your points without derogatory comments, whether they are directed at organizations or individuals.

 

[thiggins]

FYI- FCC filing posted today (06/02/2014) for the "Netgear Nighthawk X6 AC3200 Smart WiFi Router" model: R8000. FCCID: PY314200264. Internal images, and the user manual are under a 180 day confidentiality treatment

The test report shows it was was tested as a 3x3 router. So it is basically two 5 GHz and one 2.4 GHz radios in a single box.

I don't think there is any magic here with radio operation. Each will probably have its own settings in the GUI. The two 5 GHz radios at 80 MHz bandwidth each eat up all available 5 GHz channels with one left over if Ch 165 is included.

 

[jlake]

My home is not wired with cat5e. So I like the advancement in wireless technology. But honestly, once you start talking about $250-$300 for a wireless router and $250-$300 per client, it would make me re-think why I haven't gone ahead and wired it myself or apply that money towards paying an electrician to do it.

 

[azazel1024]

What we may see is link aggregation long before we see 10GbE on a router. Costs are basically firmware and that is it. As for throughput, you'll basically not see a client that can do over 1000Mbps until 802.11ax at the soonest. Even a good 3:3 client with 160MHz is unlikely to hit 1000Mbps PAYLOAD with a good connection.

Now with MU:MIMO there might be case where you have multiple clients connected where you actually do see more than 1000Mbps payload through the router...however, a pair of aggregated links and the router should handle it just fine.

I would bet quickly MU:MIMO routers will have the option early on.

 

[Razor512]

Sorry, I just really want to see some affordable 10 gigabit ethernet options. Right now many companies are price gouging when it comes to 10 gigabit ethernet. When these technologies make their way to the consumer level, the prices tend to come down by a massive amount. From a BOM cost, it should not be really expensive to produce the controllers (cost much more to make something like a low end discreet videocard). They could probably make a 4-8 port 10 gigabit ethernet AC3200 router, sell it for $200-250 and likely still make a decent profit.

I really want to upgrade, without dealing with the unreasonably high prices of many of the business level products.
overall it seems like once you start getting close to the enterprise level, companies see that as an opportunity to add a couple more zeros to the end of their profit margin
http://electronics360.globalspec.com/teardowns/information-technology/archive

 

[azazel1024]

Issue here is the 10GbE chip manufacturers are still charging a lot for the chips. I don't recall what Intel's cost is on a 10GbE chipset, but it isn't cheap. I don't think we are talking >>>$100 a pop, but it is still in the >>>$50 range IIRC and maybe over $100. That is just the chipset. I assume the switch makers are charging more than that for a 4 unit switch module that is 10GbE capable.

The switch makers have to buy their stuff from the module makers (Marvell, Broadcom, etc) before they can build their devices. So long as the module/chipset manufacturers are charging a bunch of their product, the OEM can't bring down their prices. Oh, sure, everyone is making a nice profit margin at every level on 10GbE gear, but it isn't like, Netgear, could turn around and reduce the price of an 8 port GbE switch to $100 tomorrow without taking a big loss.

I'd also like to see 10GbE ports common on routers, but you just aren't going to see it anytime soon. First off, on the router/WAN and wifi side of things, there is no consumer gear currently that justifies it. PERIOD. There are absolutely no exceptions. Even in enterprise, the use cases are on the routing/WAN side of things, not the wifi side of things.

Its going to be at least a year or two, if not more, before there are even edge use cases where wifi can push a router with only gigabit ports to the limit of what those ports can manage. On the router/WAN side of things, probably longer. Gigabit internet packages are pretty rare ANYWHERE still. Getting increasingly more common, sure...but we are still talking probably less than 1%. Packages faster than 1Gbps are I think...uh...I think NTDOCMO in Japan offers a 2Gbps consumer package. I think. I think they are the only one in the world that offers faster than 1Gbps though.

If you need a backbone faster than that for your LAN, just get a 10GbE switch and call it a day.

We are getting close, that year or two, to where AGGREGATED linking from a router would be very useful. It takes just about no hardware to do it, beyond what is already in basically all consumer routers, and just firmware to support aggregating links. That could allow you to increase links as you actually needed more speed. Considering the state of wifi, even what is likely soon, 160Mhz + MU:MIMO + 8 radio router might sometimes push a pair of aggregated gigabit links to the limit or maybe even to the point of kind of just barely needing 3 links to take full advantage of all of that wireless performance.

You still aren't going to need more than that one lowly gigabit WAN port for many years to come.

I think 10GbE being in consumer gear is still at least 2-3 years away. Its coming and fast, but still not here. It'll probably take at least 2-3 years after THAT before you might start seeing it in high end consumer routers, even though they probably aren't really going to need it to take advantage of what the router can already do (though one 10GbE would save a port or two over 2-3 aggregated GbE links).

I am curious after 10GbE if 40GbE or 100GbE will be the thing that "someday" gets pushed to the consumer market (I am sure at least 10-20 years from now). Or if something else will come along before 40/100GbE is a mainstream thing.

 

[Razor512]

Would be interesting to see some tiered implementations, e.g., sub $20 router will have 100mbit ethernet, and 30-100 will have gigabit, and 100-200 can have 10 gigabit ethernet, and probably in the $300-500 range, some 40 gigabit MMF.

A good consumer release cycle would have been,
2000-2005 1 gigabit,
2005 to 2010 10 gigabit,
2010-2015 10 gigabit with reliable teaming of multiple 10GbE ports (+ some enthusiast gear sporting some 40GbE) ,
2015-2020 40 gigabit (+ support for teaming).

By pushing things forward, the rest of the market can better innovate to develop services and improve the user experience with the added throughput.
For example, it will also make it more feasible for companies to also push compute clusters to the consumer/ home user market. (Thus also allowing them to price gouge less for the equipment)

 

[sm00thpapa]

So is the AC87U still the only router coming out with a 4x4 spacial stream?

 

[maylyn]

So is the AC87U still the only router coming out with a 4x4 spacial stream?

Yes, 10 chars

 

[azazel1024]

Would be interesting to see some tiered implementations, e.g., sub $20 router will have 100mbit ethernet, and 30-100 will have gigabit, and 100-200 can have 10 gigabit ethernet, and probably in the $300-500 range, some 40 gigabit MMF.

A good consumer release cycle would have been,
2000-2005 1 gigabit,
2005 to 2010 10 gigabit,
2010-2015 10 gigabit with reliable teaming of multiple 10GbE ports (+ some enthusiast gear sporting some 40GbE) ,
2015-2020 40 gigabit (+ support for teaming).

By pushing things forward, the rest of the market can better innovate to develop services and improve the user experience with the added throughput.
For example, it will also make it more feasible for companies to also push compute clusters to the consumer/ home user market. (Thus also allowing them to price gouge less for the equipment)

I don't disagree it would be nice to see, but both your dates and expectations aren't particularly realistic.

First, we are halfway through 2014...so unless you think that time travel is invented, why would 10GbE extend back to 2010? Especially on a consumer router. Something like that STILL isn't needed even a tiny bit on ANY consumer router. Its need on VERY high end enterprise/data center/telecom gear. I am not even aware of any routers (switches, yes, routers no) that have 40GbE or 100GbE right now.

I think having an expectation that 10GbE is going to move in to consumer routers at any price point is unrealistic before 2017 at the earliest and probably more likely 2018-2020 range. 40/100GbE is probably another decade after that.

Personally I'd expect high end consumer switches and NICs/motherboards (lets call it enthusiast) and maybe low end server/enterprise stuff to begin incorporating 10GbE ports roughly in the 2015-2017 range. Mainline consumer gear (IE a typical $80-150 motherboard, sub $100 NIC, sub $200 switch) to incorporate 10GbE more likely in the 2017-2018 range. Routers may appear with 10GbE ports roughly around then too, 2017-2018 and its darned unlikely you'll see anything on the WAN side actually NEED that for at least a few years afterward. Probably at around the same time 40/100GbE will start pushing in to the high end enterprise and data center router market as well as high end, but not obscene switches and NICs (think the situation now where a 10GbE NIC is ~$300-500 and a switch is ~$80-120 per 10GbE port). Probably take several years, maybe 3-6 after that before 40/100GbE might trickle in to the high end consumer market before getting pushed down.

That is supposing something better/nicer hasn't come along to replace ethernet by then (not that I expect it to necessarily).

It took roughly 15 years to move from 10Mbps ethernet to 100Mbps ethernet and another roughly 10 years after that to move to 1Gbps ethernet, at least in the consumer market. We've been on 1Gbps ethernet in the consumer market for about a decade now. We are relatively on the cusp of "needing" 10GbE in the consumer market with the proliferation of SSDs, faster HDDs and that sort of thing, but there still isn't a huge demand and 10GbE is still markedly more expensive and power hungry than 1GbE. It makes zero sense to drop a 10GbE NIC in to a laptop where its chipset might be burning 3-4w, plus need a heat sink, where a 1GbE chipset might be burning half a watt and not need a heat sink, plus a BOM of maybe $2-15 on the GbE chipset depending on manufacturer and capability instead of $80-200 for a 10GbE chipset, even if the laptop had an m.2 SSD in it.

As I mentioned earlier, 10GbE could be "nice" in a router, but there are zero routers that need it right now. Why invest in the power budget and expense of integrating a 10GbE swithing module in to the router when it can't utilize it at all? Its like the N150 routers with gigabit ports.

Once 160MHz 11ac comes along you could make a case that 1GbE might just not be enough. To a 3:3 client with 160MHz you might actually see in the range of 1200-1500Mbps usable payload on a very good connection, which is in excess of port speed. You'd need a 10GbE port there. You might be able to do port aggregation and monkey around with how it is establishing connections back to a server (SMB multichannel? Not sure how that might work over wifi at all) with aggregated links.

The biggest issue though is MU:MIMO, especially with 160MHz 11ac. Even with 1:1 and 2:2 clients, with 160MHz, it could be very possible with just a couple of clients, especially 2:2 ones, you might be hitting a limit with just a single 1Gbps port. There though, if you can aggregate a pair of links, you should still be radio limited and not wire limited.

Considering the cost involved in rolling port aggregation in to a consumer router (nearly free), I'd think that is the direction router manufacturers are going to go first. Also, if for no other reason, there isn't likely to be any real consumer gear with 10GbE ports (switches or NICs) for another few years. However, there is a resonable amount of gear that can do port aggregation or has NIC teams, or at least it is "low end" gear (relatively speaking) to enable that. I know personally I'd rather a high end router with port aggregation over one with 10GbE ports. Port aggregation I'd have to do nothing to take advantage of. 10GbE, I'd have to invest in a 10GbE NIC for my server at the very least and possibly a couple of 10GbE NICs and a 10GbE switch, costing me hundreds to a thousand+ just to take advantage of the router.

Even if I didn't have existing gear, buying gear that could take advantage of port aggregation would run me in the neighborhood of $70 on the low end for a semi-managed 5-8 port switch and a second NIC for a PC.

 

[htismaqe]

I am not even aware of any routers (switches, yes, routers no) that have 40GbE or 100GbE right now.

In the enterprise/telecom space, there are a few options out there that support 40GE and 100GE (e.g. Juniper M960) but once you get into devices like that, you're really starting to blur the lines between routers and switches because you're talking about multi service routers with all kinds of switching and internetworking functions.

 

[Razor512]

A bunch of companies are releasing it with a target for service providers and high end data centers, But I was more thinking more along the lines of if you build it, they will come. For example, companies like netflix would never have pushed for 4K streaming if we were still on dialup.

The idea for in home game streaming over wifi or ethernet would have likely not come up if we were still on 10 megabit ethernet.

Right now we may not have a use for 40 gigabit ethernet in the home, but if someone were to upgrade everyone to a 40 gigabit network, then probably within a few months, you will see consumer products come out that will make use of that throughput. New and highly innovative products and services can be very expensive to implement and maintain, so it is often unlikely for a new product or serve to some out before consumers have the means to actually use it. By moving technology that already exists, to the consumer level, it suddenly becomes feasible to develop the next big thing without worrying about there being an actual target market.

Back in the 90's streaming movies and TV shows was not on many peoples mind, but if at that time, someone handed me a 20mbit connection and a crunchyroll or netflix subscription, then I would have given up cable and broadcast TV many years sooner. (the need was there, but the means to acquire those services in a reasonable fashion was not available)

 

[sinshiva]

10gbe is already dead imo; it's all about USB4 and duplex Tbit WiFi; not a terrible outcome, i think. your router will do everything, or else. gamers aren't pc builders anymore. mobo makers dont seem to be trying to budget gbe interfaces fast enough. nas manufacturers dont have the power or will to take the loss required to get 10gbe to the home. so USB and WiFi already won, imo.

 

[azazel1024]

10gbe is already dead imo; it's all about USB4 and duplex Tbit WiFi; not a terrible outcome, i think. your router will do everything, or else. gamers aren't pc builders anymore. mobo makers dont seem to be trying to budget gbe interfaces fast enough. nas manufacturers dont have the power or will to take the loss required to get 10gbe to the home. so USB and WiFi already won, imo.

You are being...facecious? 10GbE is far from dead. USB4 isn't even on the roadmap, that I am aware of and it doesn't do networking. Thunderbolt networking hold some interesting potential, but the optical cable medium hasn't emerged yet, so all it allows is the equivelent of USB synching between two T-bolt devices over a pseudo networking connection.

Wireless is nice, but it still falls far short of full duplex wireless ethernet.

On 40GbE...sure, I'd imagine someone would take advantage of it eventually.

However, as it stands, current brand new SSD storage would have a hard time saturating 10GbE, let alone 40GbE. That would be like installing a light bulb years before you get electricity. Other than data center and teleco, really nothing needs or can take advantage of 40/100GbE.

10GbE would be nice to see soon. Lets get there first and worry about the next step sometime when we might be getting close to needing it, not years before anything can take advantage of it.

 

[sinshiva]

i just don't see 10gbe becoming a soho tech. 10gbe can easily be saturated with striped drives. it may not be common practice in production, but it would be more so in the home, where we'd all like to see 10GbE. it's just that wireless manufacturers are already experimenting with negotiating multiple frequencies for wireless clients and provided you aren't looking for transcoding, multiple usb3 ports will handle the majority of NAS needs. and it looks like we're not too far off from transcoding at the router anyway. and i recognize that this future leaves little room for NAS manufacturers, but if they want to stay relevant as a device, they need to make a push for 10gbe NOW. Otherwise they're all destined to be USB peripherals

 

[sfx2000]

The Asus and Netgear devices...

Well, it can be done with current chipsets and a common router core... this isn't too terribly exotic...

BOND the two 5GHz links with a client side driver to manage the client IP stack, also BOND'ed, which OSX, Linux, and perhaps Windows (never tried it, to be honest).

Bingo, you have twice the bandwidth...

A couple of comments in the thread spoke about 10G links - actually inside a SOHO router, the radio's might be a bit starved, but generally, the traffic patterns is ok for 1G Ethernet fabric.

To add a 10G switch and PHY's in a SOHO router, this would be expensive right now (right now being 060214)...

 

[hackztor]

Almost nobody at home would even need that throughput. Gig is fine for 99% out there. 10gig would probably be here if the world wasnt already moving to wifi streaming due to laptops/tablets not even using ethernet ports anymore. 10 gig ethernet just is not needed at home even in businesses its used mainly for the backbones not even the connection to the desktop. The increase in price just is not worth it for 99%. And their are 10gig switches, go buy one of those put whatever 10 gig equipment you have on that (probably one device) and enjoy the super high prices.

A bunch of companies are releasing it with a target for service providers and high end data centers, But I was more thinking more along the lines of if you build it, they will come. For example, companies like netflix would never have pushed for 4K streaming if we were still on dialup.

The idea for in home game streaming over wifi or ethernet would have likely not come up if we were still on 10 megabit ethernet.

Right now we may not have a use for 40 gigabit ethernet in the home, but if someone were to upgrade everyone to a 40 gigabit network, then probably within a few months, you will see consumer products come out that will make use of that throughput. New and highly innovative products and services can be very expensive to implement and maintain, so it is often unlikely for a new product or serve to some out before consumers have the means to actually use it. By moving technology that already exists, to the consumer level, it suddenly becomes feasible to develop the next big thing without worrying about there being an actual target market.

Back in the 90's streaming movies and TV shows was not on many peoples mind, but if at that time, someone handed me a 20mbit connection and a crunchyroll or netflix subscription, then I would have given up cable and broadcast TV many years sooner. (the need was there, but the means to acquire those services in a reasonable fashion was not available)

 

[jalyst]

but if they want to stay relevant as a device, they need to make a push for 10gbe NOW. Otherwise they're all destined to be USB peripherals

Some of the more sophisticated/popular brands do offer 10Gbe in various form-factors, QNAP as one example.

 

[azazel1024]

Some of the more sophisticated/popular brands do offer 10Gbe in various form-factors, QNAP as one example.

Still on the more expensive side though in their NAS.

10GbE is slowly moving down the pipe, but it is still going to be a few more years before it starts really taking over and replacing 1GbE.

It will come eventually. There is a need for it, even in home usage. Just not a terribly compelling reason yet. 4k video seeing wide spread introduction...yes, we'll see it more in the home. It obviously isn't needed for streaming, but with the move more and more toward PCI-e SSD storage and SATA Express, storage systems are there to fully take advantage of it and as file sizes continue to grow more and more people are going to get tired of twiddling their thumbs as they wait for their 40GB file to transfer to/from their home server or NAS because they have a 1GbE network bottle neck even though their storage can push >>>100MB/sec.

As for striped drives in the home...sure. How many people, unless striping SSDs can push >1,000MB/sec and consider 10GbE a bottle neck? I have a pair of drives in RAID0 and it can just barely make by 2x1GbE link uncomfortable. It would take at LEAST 5-6 drives in RAID0 to saturate a 10GbE. Not a very common home setup for probably anyone.

Maybe a video production environment, but not in a home.

Lots of people are wireless, but you still have to hook that wireless router to something with a wire (well, unless WISP/Cell modem). As wireless gets faster and faster, the wired link will become more and more of a bottleneck. Right now it is just extreme edge cases where GbE can be a bottleneck. That'll be less so with the introduction of 160MHz 11ac and even more so with MU:MIMO on top of 160MHz..

Aggregated links might be the solution for some, or for "ease of setup" it just might be that high end consumer routers start seeing 10GbE ports. I don't know, but I do expect manufacturers will do SOMETHING to fix the issue.