Looks like the advantages as listed are:
- Very simple visual configuration software
- Ability to view a graphical representation of device usage and prioritize a device
- Ability to update the device without interrupting the connection
- Sends setup information to Android devices via an audio tone
Very interesting. From looking at the specs, it looks like they're packaging a server and marketing it as a router. The configuration app is a very nice touch, routers have always a pain to configure. I'd love to stick this in my apartment (even though I can only receive a maximum of 10mb/s).
As a side note, I'm surprised this isn't marketed alongside the Nest branch. It really has the look and feel of Nest products with the LEDs and the speaker aesthetic. Also surprised this isn't an "Alphabet" product.
"As a side note, I'm surprised this isn't marketed alongside the Nest branch. It really has the look and feel of Nest products with the LEDs and the speaker aesthetic. Also surprised this isn't an "Alphabet" product."
It pretty much is... it's their way of sneaking the 'home automation' core/aggregation box into people's homes masked as a wireless router (for now).
I'm sure the capability you mention will come heading into the future. The only question will be : execution/security and will a competitor come up w/ a more secure/well executed product which won't serve as a data vacuum hose to google.
This kind of product launch on the heels of their massive alphabet re-org with no understanding where it came from (division wise) or what its goal is leads me to question whether the re-org is really going to move google beyond its former execution flaws ...
There are decent solutions to this already. INSTEON works great. It's got dual-band connectivity (wired and wireless) and you can either use a cloud hub if you want, or control it locally with a computer.
I can't tell you how many Insteon switches I've replaced in anger because they start flashing, stop responding to button presses, start beeping - one didn't even stop when I pulled the air gap switch. I'm well versed in their tech and in electrical systems, and everything was installed correctly, in some cases with fresh neutral wires pulled through direct from the neutral buss bar in the service panel, just to avoid potential crosstalk/current leakage, since they communicate over wires as well as via two way radio (newer spec switches).
I know it's anecdotal, but the Insteon forums are filled with similar stories. I was a very early adopter and have been through several generations of devices, and I'm committed to using another platform when I start replacing the remaining Insteon switches that will inevitably die.
I wouldn't recommend Insteon to anyone. X-10 was more reliable in my house. I'm evaluating Zigbee options now.
I just bought a Zigbee Winkhub because I was looking into something to play around with to remotely control lights. I also wanted something that I could hack and run my own code on. The Winkhub is relatively easy to hack. I've only been using mine for about a week now but am very happy with it. I've only tested mine with lights so far.
Deeper than that, even. It seems like this is intended as a local-network cloudlet[1] substrate.
Launch an app on your phone that needs a companion frontend server instance to talk to? One gets launched "in the cloud"—specifically, in a virtual cloud owned by the app author. But where is that instance, physically? Usually a provider like AWS... but with a cloudlet peering arrangement, that instance could instead end up running on your router. (Not as crazy as it sounds if your app has an N:M frontend-backend server topology.)
I've been toying with terminology to deal with the idea of a decentralized p2p cloud. Calling it a 'virtual cloud' seems off, and 'fog' has been co-opted by other kinds of tech. Any thoughts on what would be a good terminology for this?
Come to think of it, that's why Google is so behind Thread - they want a mesh networking protocol that's not exclusively local, but can connect each one of the embedded systems directly to the Internet with their own IPv6 address. Although, the protocol does support using a "gateway", too, but I have a feeling that won't be made the priority for most Thread-enabled devices.
To understand why certain products, which are seemingly unrelated to Google's core competency, have stayed with Google and haven't moved on to Alphabet, or why Google created a product such as OnHub and not Nest, you have to realize why all of these product are under Google - to track all the data that goes through them under the same generic/unified "privacy" policy, and then feed it into its advertising system.
I hate to burst your bubble, but it's actually just a branding thing, and has been in the works for much longer than anyone involved knew that Alphabet was going to happen.
The team that built this is part of the same company as Google Fiber (disclaimer: for which I work), which is in fact no longer part of Google Inc.
I've heard it's mostly a liability thing. When projects like the self-driving car were directly under Google, Google's assets were fair game in any lawsuit over that technology. With the current setup, liability is siloed in each legal entity which gives the cash cow, Google, more protection.
I wasn't commenting on why the Alphabet thing happened (I've heard all sorts of explanations but nothing official except what's been said publicly, and I don't personally have any insight). I was talking about why things in Access & Energy are still getting the "Google" label, e.g. "Google Fiber" or "OnHub by Google".
Compared to the rest of the market, paying $200 for a router isn't competitive. It just is not worth it given Google's "eh-we-will-do-it-and-stop-if-people-complain/sue" mindset.
> OnHub Wireless Router from Google and TP-LINK
> by TP-LINK
1) You complain about the quality of TP-LINK products and think the solution is....another TP-LINK product, likely measured in the same way.
2) Honestly, I cannot think of a consumer AP that would actually perform to your expectations [apartment complex with 30+ competitors, saturating all the channels] that would be priced at $300, let alone $200. I can tell you right now, this Google AP won't do it either. Its "up to 1900mbps" for a reason. Trust me. Buying this won't be a magical solution to your problems if you real world gigabit speeds.
----
To the folks complaining about "magical routers":
A) Buy something from Ubiquiti Networks.
B) Buy something you can flash with WRT firmware.
In both cases, as long as you understand the hardware specs, you'll get something better than the norm. In Ubiquiti's case, I'm 99% sure it will outperform a Google branded TP-Link AP unless Google is heavily subsidizing and QAing everything.
I have a 1 Gbps fiber connection at my apartment. I spent a good hour doing research and finally bought this http://www.amazon.com/gp/product/B00BUSDVBQ. It had 4.5 stars at the time and was the highest rated router that Amazon carried. I bought it, plugged it in and ran speedtest on it. Max speed I could get was 250mb/s. Not bad but not the gb/s that was getting advertised. Additional, over time, it would always slow down to 10-50mb/s. I called TP-Link and they said this was likely the best it'll do and the advertised speeds are in an interference free lab environment.
If OnHub fixes that second problem as they claim to, that would be more than enough reason for me to spend the $200. If they can get higher than 250mb/s speed by optimizing around the congestion, $200 will be well worth the spend.
I was in the same situation as you. Same Router too.
I could actually get close to 1 Gbps with the stock firmware. But it was unstable, requiring a reset every other day. I used an open source firmware, but it didn't contain the NAT offload module, so I would not be able to get over 200-250 Mbps. (This was is all wired, not wireless).
Since I had a wifi access point, I ended up switching to an Ubiquity Edgemax router and haven't had any problems since.
Here's the thing about TP-Link: They're a "good enough" manufactur. Meaning if you care about something, don't buy a TP-Link, but if you don't care that much and just need it to basically work "good enough" (e.g. WiFi for your mom's house) then they're fine.
I got a cheap TP-Link repeater. It works fine, nothing special, no bells or whistles. But for the price it was a good buy.
TP-Link are hardly a manufacturer of "good enough" products. Their $100 Archer C7 router is the top recommendation on TheWirecutter for over a year.[0] Their $75 Archer C5 is also the top rated AC1200 router in every single category on SmallNetBuilder.[1] It beats out more expensive models from all the bigger name manufacturers.
So what's your magical router? Does it have options actually make sense? And a help that is, like, helpful?
I'm talking about things such as "Enable XPRESS whatever". Help: "Turns XPRESS whatever on or off". And then you find out that no matter what the feature was supposed to do, theoretically, in practice it makes the router slow and unstable. Or, at worst, crashes it. No brand is immune. Many (non-enterprise) routers will advertise QoS features, but you find that you are better off without them.
That goes for almost all brands that are not classed as "Enterprise" - and even some that are.
Have you ever tried an AVM Router? Amazing things, several of their Fritz!Box lineup are actually better specs-wise than this Google router (and don’t come from a company with rather questionable update and privacy record)
It's less about your skills, and more about your presentation. "Honestly" and "Trust me" smack of salesmanship. It's a weird effect. Engineers try to be more persuasive by picking up, intentionally or not, vocabulary used for persuasion. Unfortunately, there is a lot of nuance that's hard to control.
Tighten up your writing, and perhaps speech, so you simply state the facts as you see them. This router will meet your needs because it has these 3 features. 1,2,3. It's much harder to argue with flat statements of fact as opposed to the salesmanship trick of, i would lie to most people but i like you so i'm going to tell you a secret.
Or, you know, don't. If it's not a problem for you don't waste time on it, but that's the underlying effect.
I think the underlying problem is my incentive is for people to disbelieve me the first time I tell them something. I don't really care if they believe me since after I am proven right I get more to fix the problem. [e.g. The last time, I got ~$6k] :/
So I'm not really motivated to fix it, y'know?
I probably should fix it since I'd be a better person for it but my financial incentive isn't aligned with fixing it.
My point: you're a random stranger saying to another random stranger "trust me" without justifying your request to trust you. Why should I trust you?
Your reply: After I've convinced them that I know enough to fix the problem (by either demonstrating my expertise/qualifications/experience or getting someone else to vouch for me), they believe me.
Well duh! If you demonstrate why they should trust you, they'll trust you. They won't just randomly pay you money to fix shit if they don't trust you.
No experience with Draytek's routers, but I've been happy with their ADSL/VDSL modems (Vigor 120/130). They have some really nice features, like PPPoA -> PPPoE passthrough.
I've never had problems with my Router. I do however have problems with Google ChromeCast vanishing from other devices -- which according to Google is a problem with the router (although for me the problem mostly occurs when the ChromeCast decides to use the same WiFi channel for its internal WLAN as the WiFi it's registered on).
I also have occasional problems with Linux devices being picky and flaky -- sometimes not connecting at all (repeating the "connecting..." phase ad nauseam) and sometimes just dropping the connection until I restart the device.
I find it dubious that these problems would be fixed by a Google branded router.
I have a Ubiquiti AC access point at home and the only thing I could see the OnHub doing better is the auto-sensing other networks and working around them. I'm not sure how that would work yet since changing channels would kick everyone off wifi for a second but the Ubiquiti "auto" channel is only changed/detected at boot (probably for the reason I just mentioned).
Not really. The Linksys WRT1900AC has a 1.2ghz dual core CPU as well.
At some point you need a general purpose CPU to run everything and it's not hard to just throw an ARM SoC at it which is going to be in the 1-1.5ghz range and dual core because that's just what low-end is now.
The CPU is inline with top of the line routers these days (e.g. Asus RT-AC87U is a 1Ghz dual core). 1+Ghz dual core CPUs are really cheap these days.
The RAM is about 4x what other routers are packing but the cost difference between a 1Gbit chip and 4Gbit chip is a few dollars. Prices have been plummeting this year and 8Gbit chips replaced 4Gbit as the new hotness. Not to mention DDR4 rolling out so there might just be dumping going on that makes adding 1GB of RAM a non issue.
There's a specs link at the top of the page, but I'm also not sure what parent poster means. Just confused, I think.
The specifications are on a par with low-end consumer-grade NAS products, and it has a USB 3 port, but nothing on the site says anything about providing storage or other services, except for router-type and supporting configuration services (I suppose "keep an eye on your network" is a bit of an extra).
"Plenty of room, OnHub has 4GB of storage space, so there’s plenty of room for auto-updates and the latest software features" is hardly selling a server. All it claims to be is a fancy router.
> The configuration app is a very nice touch, routers have always a pain to configure.
Basically this boils down to:
The average home router which has basic to no settings (many even lack channel selection) but usage is poor to suboptimal. I had a Netgear router which had broken internal static routing - and netgear's response was to buy a different router (even though you could set it inside the router...).
vs
Buying a higher power router or installing DD-WRT, Tomato or any of the other firmwares - now you have 1000 configuration items but you don't know which ones to tweak to give you good performance.
I believe most people fall into the middle - they want something that's easy to use with good performance but with the ability to adjust the 1000 configuration items if they wanted to.
> I believe most people fall into the middle - they want something that's easy to use with good performance but with the ability to adjust the 1000 configuration items if they wanted to.
I think most people want something that just works. (Most HN readers might want something where they can adjust 1000 configuration settings, but most people I think would never want to.)
> I'm surprised this isn't marketed alongside the Nest branch.
I'm surprised that it isn't a Nest branded product that the Nest division ultimately controls. Looking at Google's history, I wouldn't be surprised if the Nest division came out with their own wifi router as well.
I'm hoping that this doesn't turn into another Google TV release where the product is barely beta.
I think the best way to judge this product is as a hub for a smart home. The fact that it is an easy to wifi router is incidental (in spite of the fact that is all the marketing is talking about). It seems pretty clear, "onHub" is suppose to be a hub for a bunch of "on" branded smart home devices.
2.4 GHz wireless 802.11b/g/n 3x3 with smart antenna
5 GHz wireless 802.11a/n/ac 3x3 with smart antenna
AUX wireless 802.11a/b/g/n/ac 1x1
Ethernet switch QCA8337 Gigabit sw
WAN port 1x 10/100/1000 Mbps
LAN port 1x 10/100/1000 Mbps
$200 for the unit -- so it's competing on the higher end of consumer routers against offerings such as Apple's Airport. They seemed to have paid a lot of attention to human-friendly UI/UX though.
My point was simply a comparison, yes I should have provided my depth. But at $200 for a premium consumer grade router you are definitely within the realm of a nice pfsense (or alternative) system.
You could, but at that price it would most likely be a full desktop running pfsense, and who wants that kind of power draw at home? Comparable low-power pfsense boxes are about the same price, or even more for the ones advertised on the pfsense website[1].
The OnHub spec page mentions a draw of <36W[2], which is more than I would have guessed, but it's still less than most desktops. For a device running 24/7 in a residential setting, it makes sense to be low-power.
I wouldn't really call it a full desktop. Intel DQ77KB mini-itx, lga1155 with dual Intel nics is what we use for $100 off Amazon. Pair it with a cheap Celeron and some used laptop memory off ebay. At the wall ~15W.
Right now I have one of them with pfsense and its openvpn server running. But that one was upgraded to an Intel i5-3470T so it pulls about ~25W. We went with that cpu for it's AES stuff but it turns out that pfsense (or maybe its freebsd, not sure) can't completely take advantage of it. But otherwise it works great for a vpn that we use when we're outside and on mobile.
Another with a Celeron G1620 is running as an vsphere server. I really love the board.
came here to comment on the UX. It makes sense, an average person should be able to detect if someone is connected to their wifi when no one should without knowing what DHCP is.
Hard to believe there's a huge market for $199 WiFi routers, no matter how sophisticated.
On the other hand, even a sprinkling of installations is probably pretty valuable if it allows Google to derive more accurate ISP performance numbers. There's got to be some benefit in having thousands of devices directly connected to the ISP's network on the consumer end 24/7.
Apple's Airport Extreme sells for the same price and supports AC. I wouldn't expect Google to try to sell something more expensive without Apple's care for details.
Obligatory "buy refurbished equipment from Apple" comment (full manufacturer's warranty; I've never had any problems). The refurbished Airport Extreme is $129, which seems like a pretty good deal. Try poking around under "Mac Accessories".
I spent $150 each or something like that on my apple routers about 5 years ago, they both are still running strong but since technology has of course advanced so much in the past few years, I would spend that on a router with the hope it lasts the same time.
I paid almost that for my current router[0]. I think mine has better specs, even not being an AC (few, if none, of my devices use it anyway). Google's hub has no USB ports, no guest wifi (from what I can gather). Mine, however, has those things, though to be honest, most of the time, the printer or USB storage doesn't quite work right, though that could because I use a Mac and the driver's aren't supported as well.
If you're not spending at least $100 on a wireless router, you're probably giving up significant amounts of the performance you're paying for from your ISP. Routers are computers. The $40 units just can't handle modern usage patterns (multiple phones / tablets / computers with relatively high speed caps) very well.
It's my single biggest recommendation when people complain about the speed of the Internet in their home. More often than not, they're underspending on their networking equipment, not their ISP package.
> More often than not, they're underspending on their networking equipment, not their ISP package.
Often, their networking equipment is part of the ISP package because the ISP wants to provide a seamless experience, and doesn't even provide easy access to the information necessary to use your own equipment.
I don't even know what this means. Are you saying that your ISP doesn't allow you to plug standard ethernet devices into their lines? I own my own modem and router and do what I want with them. I don't follow you.
> Are you saying that your ISP doesn't allow you to plug standard ethernet devices into their lines?
Using your own router rather than the ISPs preconfigured router requires more than just plugging a personally-purchased router into their line (even in the simple case where you aren't using the ISPs bundled cable/phone/internet service that's all delivered through a common system.)
At least in the simple case, someone with modest technical skill can figure out how to get that information relatively easily, but most ISP customers aren't even modestly technical.
Since when? When I got my new router, I plugged it in and was online. I did some configuration (and flashing) afterwards, but it certainly worked when I plugged it in. Isn't that the point of consumer products?
When I've had a cable modem, all I've had to do is connect my router to the modem via Ethernet. When I've had modemless service via Ethernet jack, all I've had to do is connect my router to the wall Ethernet jack.
So what are you talking about? Purchasing your own cable modem? Or something else that I've never had to do?
If you have DSL, the modem and the router are the same box, so if you want your own, you have to throw out the one the ISP gave you, and configure yours by yourself. In practice, I bet you could just call up the ISP if you're having trouble, but I agree that an ISP has an incentive to give you a good unit to start with.
I haven't had DSL in a while, but when I did, the ISP just provided a modem with a phone line connector and a single ethernet connector. There's no requirement for DSL to bundle the modem and router.
Sure, lots of DSL ISPs used to do that in the old days before WiFi was a common consumer demand, and some still might.
Now, the ISPs that are using DSL (e.g., AT&T U-Verse which uses VDSL2 where it isn't FTTP) seem (from what I've experienced and seen with other people that have DSL-based service) to deploy all-in-one modem/routers (and, when they have services like TV and Phone as well as basic internet, to run them through the supplied all-in-one box, as well.)
> There's no requirement for DSL to bundle the modem and router.
Sure, its not a theoretical requirement, just practically what ISPs seem to do now.
Gotcha, I thought it was being presented as some sort of technical constraint.
An interesting variant on all of this is FiOS, which is often brought into the house on coax which plugs into their router. It's possible to buy something that will bridge the coax to ethernet, but they're pretty hard to find.
I think this might be going away as they increase speeds. I originally had the coax, but then they replaced it with ethernet coming in to the house. I kept their router anyway because it actually seems to be pretty good.
> If you have DSL, the modem and the router are the same box, so if you want your own, you have to throw out the one the ISP gave you, and configure yours by yourself.
The usual recommended approach I've seen is to keep the ISP-provided modem/router, plug your router into it, and make certain configuration changes on one or both of the devices (though, for the same ISP, I've seen conflicting information on the correct configuration changes, with clearly dated assumptions, but still possibly correct as far as it goes information, in the official information from the ISP.)
But my basic point is that for most home users, they aren't conceptualizing it as spending separately on "internet service package" and "network hardware", the network hardware is part of the internet service package.
Most modem-router combos support "bridge mode" which allows you to connect your own router's "Internet/WAN" port to one of the the modem-router's ethernet ports and have it initiate PPPoE to your ISP.
If that doesn't work, you can also just connect the modem-router's ethernet port to a LAN port on your own router, and just disable DHCP/DNS on the router.
My experience is that the DSL modems only contain mid-range routing and wi-fi equipment (10/100 switches instead of Gigabit, 802.11 b/g/n that run with one antenna instead of six and only on 2.4 GHz band). It's adequate to get things "working" but if you pay for a faster VDSL plan (usually 25 or 50 Mbps packages) you can see improved Wi-Fi performance with the right third-party router and laptop/phone.
My ISP provides an integrated modem/wifi router and they don't provide a bridge mode. I've long given up trying to fight against that with my own equipment.
It's not even a battle worth fighting; your average consumer router is significantly less powerful than what the cable company gives you. Even if you do get it in bridge mode, you'd need something more than a $200 consumer router to get the max speed out of it.
So not true, the RT-AC68U will walk circles around any eMTA an ISP will give you. The eMTA's cable companies give are expensive because a DOCSIS 3.0 modem is already $80 on average, adding a VoIP gateway and WLAN router plus a sufficient profit margin adds up.
Well, basically, a $200 home router will peak well before maxing out a 150 mbps connection unless you turn off all the QoS, UPnP and security features... but at that point you're better off just using the eMTA (which can handle the speed because it never had any of those features in the first place).
Once again, not true. NAT and QoS are the most expensive tasks a consumer router will perform, and even at 1Gbps it's rather mundane work even for a slower CPU (the 600MHz CPU in the RT-AC66U is sufficient).
Once you start looking at the latency requirements to achieve 10GBps interconnects you're going to have issues with what you can reasonably expect from the $200 price range, but consumer networks have a long ways to go before that's going to even need to be thought about.
I think you're overestimating the price of hardware required to handle 100 mbps. Broadcom asics in datacenter switches handle on the order of 100gbps and they are on the order of $100.
That's about what I paid for my ASUS RT-AC68P. I went through a couple different sub-$100 routers (Belkin and Linksys) before settling on this one. I've not yet had to reboot it for any reason other than configuration changes, something I definitely couldn't say about the others. Previously I was using a Buffalo Wireless G router with DD-WRT, but it finally gave up the ghost after about 7 or 8 years of service. Having built in samba / FTP / DLNA support of USB drives makes it worth the added expense, as does QoS and SSH access. I'd highly recommend it.
I bought some 802.11AC router from Asus a few months ago for around that much, and I was absolutely floored at how much better it was than my old WRT54GS that I had been using with DD-WRT for years. Now I'm easily able to download consistently at 16MB/s from the internet, and do large file transfers at even higher rates between devices at home. I've mostly stopped using gigabit ethernet at home even because it's so fast.
Agreed. It looks a little like a squatter Echo. Cylindrical, ring of light at the top. But I suppose Amazon will avoid any legal action since it's pretty tough to win a court case with the argument "they copied my cylinder." Hopefully the different functionality of each device will keep any ugly legal battles from ensuing.
But with the a computer embedded in this router along with a speaker and bluetooth and zigbee radios this could be customized to offer a lot of the features that the Echo offers (minus the voice control). I hope that doesn't cause any problems.
About 80% of the space in the Amazon Echo is taken up by the speaker. The Google router just has a little 3w speaker in it, so I don't think Google is gonna be converting it into an entertainment device anytime soon.
Hopefully this device actually has the CPU power to run QOS at more than 50mbps... the same can't be said of any other consumer router in the $200 price range.
Quality of Service is when the router shapes the traffic making sure all devices on the network get their 'fair' share so to speak. It can prioritize traffic depending on application and use. So you could deprioritize streaming video for instance if you want to make sure that your skype call went through.
This takes some processing power which not all routers have a lot of.
> This takes some processing power which not all routers have a lot of.
I've set up and hand rolled my own Linux-based routers in the past on cheap leftover hardware (first gen Celeron). Adding QoS introduced literally zero noticeable CPU overhead.
I'm not saying you're wrong but you're clearly overselling the computational cost of this feature.
Fair enough! I've never really delved too deeply into it as most of the time the stuff I deal with is pretty simple. Though some other comments parallel to mine seem to agree with me
My router is a Asus RT-N16. It has a 480 MHz Broadcom BCM4716 chip. It uses 100% of the CPU just do to NAT on my 100 mbps connection. Doing QoS on it is out of the question. As is doing VPN at those speeds.
You are right. It isn't that much CPU, but with modern connection speeds (>=100 mbit) you actually need a half way decent CPU. Your old Celeron was probably routing a < 10 mbps connection.
You're right about a few things, but I don't think you're painting a true picture of things.
Yes, your router uses much CPU for NAT. That's because it is a complex process which requires mangling the packets and keeping lots of state in a big table.
QoS however is relatively cheap. It's merely looking at fixed offsets in the IP packets and deciding which of those (mangled) packets must go now and which can go later. It requires some extra memory for some extra packet-queues, and a lookup in one of iptables' smallest rulesets. That's about it.
Yes, I routed things at slower speeds, but you got to have a perspective on things: You must NAT all the traffic, but only need to QoS the outgoing traffic.
This means that if the downstream speeds have increased 100x recent years while upbound speeds only 10x (which is my fair assesment of the market), your NAT cost is now 110x higher, while the QoS cost is merely 10x.
And QoS was the cheap load. If your router is already buried in other work, I'm sure QoS will feel extra painful, but so would anything else. On a healthy load you probably wont see it in the load-graphs at all.
Also worth keeping in mind is that QoS as a feature has diminishing returns the higher upload-rates you have. If you have 128kbps upload, making sure the right packets get out first is excessively important. If you have Gigabit upload, not only will QoS be more expensive, but you are almost guaranteed to have no benefit from it what so ever.
So yeah, I stand by my argument. QoS is not a heavy feature, even though marketed as such.
QOS is quality of service. It lets you prioritize certain traffic over others. So if I'm running BitTorrent but trying to watch Netflix, I can set up a QOS rule to prioritize Netflix over the BitTorrent. Netflix might only need 4Mbps but a torrent can take 100% of my connection. So I would set a QOS rule to make sure that Netflix gets what it needs then the torrent can have the rest.
Inspecting every packet as it runs through the network takes a lot of processing power. Many router manufacturers put just enough hardware in the box to handle fast connections, but not enough to handle complicated QOS rules on top of that.
Looks like Google is doing the same; that CPU probably isn't fast enough to handle NAT and QoS at the speeds the WiFi radio is capable of; much less a gigabit Google Fiber connection. While I understand consumers aren't going to spend $500 on a router, it's a bit disappointing to see Google engage in the same behavior that Linksys, D-link, et al do.
The specs are on the OnHub page at the bottom :) Looks like it's using a Qualcomm IPQ8064; which is a dual-core ARM design. I don't know what other devices use this CPU so I can't compare, but if I had to guess this device probably tops out around 100mbps if NAT and QoS are turned on.
Just interesting to note that the form factor is quite similar to Amazon Echo (slim cylindrical tube), although the function is completely different (router vs "personal assistant device").
It has a speaker built in. So they'll have that with Google Cast Audio, Google Now, etc.
I am pretty sure they have a taken a leaf or two out of Amazons book.
True, although the audio engineering on the Echo appears to have been quite purposeful. Not sure what kind of performance they'll get out of the speaker on this Google device. But the functionality may well be there, I agree.
I presume they're designing something to sit visibly on a shelf, but I quite like having things in a more stackable form factor. e.g., an external 3.5" on top of a Mac Mini sitting on top of a four-bay NAS, etc.
I guess that creates issues with interference with the antenna functioning though?
Can anyone explain why a Router needs 4GB of storage ?
What do they mean by "OnHub has 4GB of storage space, so there’s plenty of room for auto-updates and the latest software features." ? Is software for a router that large ?
The Apple TV has 8GB of storage, so when you start streaming a TV show or movie it begins downloading the whole thing so internet interruptions don't mess up your viewing.
The pure router parts? No. But stuff like the home automation features I can see why they'd fit more storage, especially in early versions where they don't know quite yet what they'll want to put in later. (Of course, swappable storage would be even better for that, but has other issues)
A local cache is of course useful for streaming video, but with home automation it can also be useful in the reverse direction. If your Nest cameras are not getting enough bandwidth to send up full 1080p back to the service, they can send 480p up to Google, cache the full-res video on the OnHub router, and upload the buffer late at night when bandwidth is more available.
OnHub looks nice, but honestly the Apple AirPort Time Capsule is a great device, and also gives you Time Machine backups. The only feature I've found missing, is VPN site-to-site and VPN client connections.
Coolest thing I see is on the NewEgg page where it says it's "weave" ready, and you can see the Zigbee/Bluetooth radios. It's a router, and it's the smart home hub.
I'm really surprised this part didn't generate more attention from the HN crowd. Currently IoT devices always end up having some kind of protocol adaptor unit, like wifi->zigbee and it sucks!
As far as I know, the Almond+ is the only other router with 802.15.4 support.
1. Overall-- this can, theoretically speaking, integrate all- WiFi IoT, Zigbee IoT, Thread IoT, Weave IoT. Whichever way the market picks-up, Google can.
2. On top of that, the specs are good enough to stream multiple IP cams.
3. Shot in the dark, but with all the "six 2.4GHz antennas, six 5GHz antennas, and a congestion-sensing antenna" they're probably trying to fix the classical WiFi congestion and star-topology range issues that comes with WiFi IoT. Especially because it supports 128 devices in WiFi.
4. Rumors are, it already runs Brillo, their IoT OS they showcased at IO.
5. Its bridge mode support be something they can use to have "multiple BRs" as promised by Thread?
The only privacy link I could find on the page was Google's generic privacy policy. Is there more information somewhere?
With its "advanced and always-evolving security features", I'm wondering whether and what information Google would be gathering about my local network and my network traffic.
"You agree that Google may collect and use technical and related information, including but not limited to information about your computer and/or mobile device, operating system, peripherals, applications, connected devices, network traffic, and data use to facilitate the provision of the Software and Services, including support and other related services."
It goes on to state "The OnHub Privacy FAQ describes the categories of data collected and how you can use privacy settings to change what data is collected by the Services."
Unclear why they refer to it as "The OnHub Privacy FAQ" (implying proper noun) but then the page is called "OnHub, the Google On app and your privacy" and it's not structured as an FAQ. I can't find anything else it could refer to, though.
Is it weird that I assume that both Google and NSA are recording all of my digital actions? Does no one else assume this? Which group is more delusional?
Here in CH virtually everyone receives a modem/router combo for free from their provider (ie. they borrow it to you). They are also usually preconfigured - so you just need to plug them in. Is it the same in the US? I guess at a 199$ price point this will be a tough sell.
Yes, it's the same in the US. ISPs provide a modem/router combo and charge like $7 a month for it. The disadvantage is that the router portion of the unit is usually very poor. AT&T is still giving out modem/router combos that us 802.11b/g, not even n, much less ac. Purchasing a separate router lets you use better wifi technology as well as enable useful features like QoS settings that aren't available on the ISP-provided hardware.
Here in Germany those offers are similar. And nearly everybody somewhat tech-savvy doesn't get or use them, because most of the time they are pretty crappy. 150€+ routers are a limited market though, I've mostly seen top fritzbox models that come with DECT phone support etc.
Here in Singapore, some of the providers offer free but crappy router for their 500mbps and up. While other ISP offers S$50 voucher for the routers they sell.
I imagine it's like Echo, and the various sticks out now; if not on Android, you connect to it from your phone as a wifi access point, and set it up from there. When done it reboots and becomes part of your regular network.
So what does this device primarily do that I need right now? Just throttling internet for certain devices and prioritizing others? I'm not even entirely sure what 'uses smart software to find the best WiFi connection' means. What's the value prop here given the device costs $200? It's not like it can better your internet somehow.
"any plastic router has some sort of QoS build in"... that doesn't work.
If you don't have an expensive product, do some tests with and without the router's QoS. You'd be better of with a raspberry Pi doing QoS in most cases.
The only way I can interpret that quote is, "it can automatically switch wifi radio channels when the one its using gets congested". I don't know, maybe there's more to it than that.
With six 2.4 and 5 GHz antennas each, that's two antennas times three (non-overlapping) channels simultaneously in each band. Assuming it can somehow get device A to communicate on channel A and device B to communicate on channel C, etc., you could avoid some congestion while still allowing all those devices to maximize the available bandwidth.
No $40 routers that I know of do anything like that.
meaning that if everyone in the neighborhood owns one, there will be six times as much congestion and everyone's internet access will be even slower than before!
I'm not 100% this is true, but with multiple antennas, you can do beam-forming. The basic idea is that you can use interference to point the wifi signal at devices that are using it heavily at the moment, boosting their signal and thus their speed.
When I upgraded from a generic $50 linksys router to a $200 Apple Extreme it vastly increased the range and speed of the wireless. Before it my TV's netflix connection would drop all the time or switch to low-res mode.
So it's conceivable that it would better your Internet.
That's interesting. The largest apt I've lived in was a 850 sq ft 2br so I've never had any issue with the range or lag from my router. Thanks for sharing.
I live in a 200 sq ft studio and I've had issues with interference with 2.4 GHz Ubiquity equipment and various other routers (probably due to having too many neighbors within range) -- switching to a $200 Apple Airport Extreme with "more antennas" made most of my problems go away.
However, seeing better performance 5 GHz 802.11 n/ac requires you to have both a good router and a good laptop -- my experience is that unless your users/family blindly uses Macbooks, they won't see any improvement from a $200 router because their 2-3 year old $500 Windows laptop only has a 802.11 b/g/n 2.4 GHz-only 1x1.
Most of this is because of the terrible Linksys software. I've had routers that acted terribly, and seen them perform much better with DDWRT installed.
It has a 3rd radio that monitors the channels you aren't currently using to see if any of them are better. The first two radios in the device can't switch without dropping connections, so they can't go hunting for new channels once the AP is turned on.
My Airport Extreme chooses channels automatically at boot time, then never again. This is fine for my house though, because my neighbors are reasonably far away.
At work, our Meraki AP has this kind of 3rd radio, and we get very good performance in a terrible RF environment, where lots of other equipment (from Apple and Ubiquiti) has failed.
The Fritz!Box 7490 has 3 radios, for exactly this purpose. That’s how it supports automated channel search without disconnecting users, and how it is able to provide 1.3Gbps via WiFi.
If it does have 3 radios, that's awesome. They should make that much more explicit. I've looked all over their site, and I can't find anything that confirms this. Everything on their website just references plain old simultaneous dual-band with MIMO antennas.
For me, at least, a 3rd radio will be on the list for my next AP, so I'll take a close look at this one, too.
Well, they claimed it. I am not sure if they do have one, but switching channels and empty-channel-autodiscovery without disconnecting is a feature in the version that I have.
The issue is just that AVM produces different versions of the Fritz!Box for every ISP, so some have more storage, or more radios than others.
That Fritz!Box 7490 is US $320 on Amazon. OnHub is $200. I doubt many the sub $100 routers most people get do that sort of thing. So maybe nothing special but that is quite a high end router you're comparing it to.
Doesn't this only really buy you time until everyone around you has the same router? If everyone is constantly changing channels to compensate for neighbours I can't imagine the performance to be great, but perhaps I'm missing something.
The thing that throws me here that others haven't commented on yet... it's built 'in partnership with TP-Link'. TP-Link makes budget hardware, that doesn't last much longer than it's one year warranty.
Charging $199 for a TP-Link device sounds... overpriced. If this had been built by ASUS, then maybe they'd have something.
> Google’s hardware partner for the initial OnHub is TP-LINK; it plans future devices with other manufacturers as well, including one from ASUS later this year.
That may be your experience, but I've been using TP-Link as my go to access points for myself and my clients (home and small business) for the past 6 or so years. I've deployed maybe 20 APs in that time and none have failed yet.
Not communicating by blinking lights is not a feature to me. Progressive enhancement all you want, have a nice app for involved communication, but keep some LEDs for core status please.
From the design goals outlined in the FAQ, it seems clear to me that Google expects there to be a great deal more bandwidth hungry devices in the home speaking first to each other, and then out to the WAN, in the near term.
Including Bluetooth, Weave, and Thread, along with up to 128 devices, gives me a rough sense of the scale of connected devices they anticipate in each home. Given that it also has a reasonable compute capabilities and cooling as one of the highlights, I think it's far to say that it could do a lot of processing on its own before reaching back home to the GOOG DCs.
640 comments
[ 4.1 ms ] story [ 319 ms ] threadAs a side note, I'm surprised this isn't marketed alongside the Nest branch. It really has the look and feel of Nest products with the LEDs and the speaker aesthetic. Also surprised this isn't an "Alphabet" product.
It pretty much is... it's their way of sneaking the 'home automation' core/aggregation box into people's homes masked as a wireless router (for now).
I'm sure the capability you mention will come heading into the future. The only question will be : execution/security and will a competitor come up w/ a more secure/well executed product which won't serve as a data vacuum hose to google.
This kind of product launch on the heels of their massive alphabet re-org with no understanding where it came from (division wise) or what its goal is leads me to question whether the re-org is really going to move google beyond its former execution flaws ...
I know it's anecdotal, but the Insteon forums are filled with similar stories. I was a very early adopter and have been through several generations of devices, and I'm committed to using another platform when I start replacing the remaining Insteon switches that will inevitably die.
I wouldn't recommend Insteon to anyone. X-10 was more reliable in my house. I'm evaluating Zigbee options now.
Revolv Hub supported devices:
http://support.revolv.com/knowledgebase/articles/329116
Deeper than that, even. It seems like this is intended as a local-network cloudlet[1] substrate.
Launch an app on your phone that needs a companion frontend server instance to talk to? One gets launched "in the cloud"—specifically, in a virtual cloud owned by the app author. But where is that instance, physically? Usually a provider like AWS... but with a cloudlet peering arrangement, that instance could instead end up running on your router. (Not as crazy as it sounds if your app has an N:M frontend-backend server topology.)
[1] https://github.com/cmusatyalab/elijah-cloudlet
* It’s a play on the word “shard”.
* It suggests privacy and protection.
* It suggests something that is nearby, as opposed to “the cloud” that is far away.
…
Unfortunately, the word is probably too susceptible to negative connotation.
That's why OnHub is a Google product.
The team that built this is part of the same company as Google Fiber (disclaimer: for which I work), which is in fact no longer part of Google Inc.
I've heard it's mostly a liability thing. When projects like the self-driving car were directly under Google, Google's assets were fair game in any lawsuit over that technology. With the current setup, liability is siloed in each legal entity which gives the cash cow, Google, more protection.
More here: http://www.theverge.com/2015/8/13/9149431/alphabet-google-co...
Here is the help page on the topic of data collection and privacy: https://support.google.com/onhub/answer/6246642?hl=en&ref_to.... And see also, how to set a whole bunch of privacy settings: https://support.google.com/onhub/answer/6279845?vid=1-635755.... Disclaimer: I work on the project. I am not a spokesperson or legal or marketing, or anything like that.
Eh? You must buy some shitty routers, no offense.
Compared to the rest of the market, paying $200 for a router isn't competitive. It just is not worth it given Google's "eh-we-will-do-it-and-stop-if-people-complain/sue" mindset.
----
To Melvin, since I'm submitting to fast:
http://www.amazon.com/dp/B013ALA9LA?ref=spks_0_0_2170510902&...
> OnHub Wireless Router from Google and TP-LINK > by TP-LINK
1) You complain about the quality of TP-LINK products and think the solution is....another TP-LINK product, likely measured in the same way.
2) Honestly, I cannot think of a consumer AP that would actually perform to your expectations [apartment complex with 30+ competitors, saturating all the channels] that would be priced at $300, let alone $200. I can tell you right now, this Google AP won't do it either. Its "up to 1900mbps" for a reason. Trust me. Buying this won't be a magical solution to your problems if you real world gigabit speeds.
----
To the folks complaining about "magical routers":
A) Buy something from Ubiquiti Networks.
B) Buy something you can flash with WRT firmware.
In both cases, as long as you understand the hardware specs, you'll get something better than the norm. In Ubiquiti's case, I'm 99% sure it will outperform a Google branded TP-Link AP unless Google is heavily subsidizing and QAing everything.
If OnHub fixes that second problem as they claim to, that would be more than enough reason for me to spend the $200. If they can get higher than 250mb/s speed by optimizing around the congestion, $200 will be well worth the spend.
I could actually get close to 1 Gbps with the stock firmware. But it was unstable, requiring a reset every other day. I used an open source firmware, but it didn't contain the NAT offload module, so I would not be able to get over 200-250 Mbps. (This was is all wired, not wireless).
Since I had a wifi access point, I ended up switching to an Ubiquity Edgemax router and haven't had any problems since.
Not really a fan of TP-Link.
I got a cheap TP-Link repeater. It works fine, nothing special, no bells or whistles. But for the price it was a good buy.
[0] http://thewirecutter.com/reviews/best-wi-fi-router/
[1] http://www.smallnetbuilder.com/tools/rankers/router/result/1...
I'm very happy considering the price and openwrt support.
I'm talking about things such as "Enable XPRESS whatever". Help: "Turns XPRESS whatever on or off". And then you find out that no matter what the feature was supposed to do, theoretically, in practice it makes the router slow and unstable. Or, at worst, crashes it. No brand is immune. Many (non-enterprise) routers will advertise QoS features, but you find that you are better off without them.
That goes for almost all brands that are not classed as "Enterprise" - and even some that are.
I don't trust people who say "Trust me." :)
Tighten up your writing, and perhaps speech, so you simply state the facts as you see them. This router will meet your needs because it has these 3 features. 1,2,3. It's much harder to argue with flat statements of fact as opposed to the salesmanship trick of, i would lie to most people but i like you so i'm going to tell you a secret.
Or, you know, don't. If it's not a problem for you don't waste time on it, but that's the underlying effect.
I think the underlying problem is my incentive is for people to disbelieve me the first time I tell them something. I don't really care if they believe me since after I am proven right I get more to fix the problem. [e.g. The last time, I got ~$6k] :/
So I'm not really motivated to fix it, y'know?
I probably should fix it since I'd be a better person for it but my financial incentive isn't aligned with fixing it.
Your reply: After I've convinced them that I know enough to fix the problem (by either demonstrating my expertise/qualifications/experience or getting someone else to vouch for me), they believe me.
Well duh! If you demonstrate why they should trust you, they'll trust you. They won't just randomly pay you money to fix shit if they don't trust you.
But hey, go keep trusting the people who already screwed you.
1) You want to fix a shitty TP-Link product with another TP-Link product.
2) How is that not that?
Now I'm on fibre I'm using the Draytek purely as an access point, and BT's modem to connect. The Draytek'll need replacing soon.
I also have occasional problems with Linux devices being picky and flaky -- sometimes not connecting at all (repeating the "connecting..." phase ad nauseam) and sometimes just dropping the connection until I restart the device.
I find it dubious that these problems would be fixed by a Google branded router.
The detailed specs page shows a 1.4 GHz Dual Core processor and a 1GB of DDR3 RAM. Thats a little ridiculous for a router
At some point you need a general purpose CPU to run everything and it's not hard to just throw an ARM SoC at it which is going to be in the 1-1.5ghz range and dual core because that's just what low-end is now.
Agreed, the specs between the WRT1900AC and the OnHub are strikingly similar. Even the advertised smart features.
The RAM is about 4x what other routers are packing but the cost difference between a 1Gbit chip and 4Gbit chip is a few dollars. Prices have been plummeting this year and 8Gbit chips replaced 4Gbit as the new hotness. Not to mention DDR4 rolling out so there might just be dumping going on that makes adding 1GB of RAM a non issue.
The specifications are on a par with low-end consumer-grade NAS products, and it has a USB 3 port, but nothing on the site says anything about providing storage or other services, except for router-type and supporting configuration services (I suppose "keep an eye on your network" is a bit of an extra).
"Plenty of room, OnHub has 4GB of storage space, so there’s plenty of room for auto-updates and the latest software features" is hardly selling a server. All it claims to be is a fancy router.
Alphabet has been announced, but AFAIK the actual corporate transition hasn't yet occurred.
Basically this boils down to:
The average home router which has basic to no settings (many even lack channel selection) but usage is poor to suboptimal. I had a Netgear router which had broken internal static routing - and netgear's response was to buy a different router (even though you could set it inside the router...).
vs
Buying a higher power router or installing DD-WRT, Tomato or any of the other firmwares - now you have 1000 configuration items but you don't know which ones to tweak to give you good performance.
I believe most people fall into the middle - they want something that's easy to use with good performance but with the ability to adjust the 1000 configuration items if they wanted to.
I think most people want something that just works. (Most HN readers might want something where they can adjust 1000 configuration settings, but most people I think would never want to.)
I'm surprised that it isn't a Nest branded product that the Nest division ultimately controls. Looking at Google's history, I wouldn't be surprised if the Nest division came out with their own wifi router as well.
I'm hoping that this doesn't turn into another Google TV release where the product is barely beta.
Edit: I spoke too soon.
One LAN, one WAN.
Networking
Would probably wait for reviews to see what's up.
The OnHub spec page mentions a draw of <36W[2], which is more than I would have guessed, but it's still less than most desktops. For a device running 24/7 in a residential setting, it makes sense to be low-power.
1. http://store.pfsense.org/SG-2220/
2. https://on.google.com/hub/#specs
With a case, power supply, CPU, heatsink, and RAM the price starts to approach the OnHub though, without any Wi-Fi nics.
The DQ77KB is a cool little board. Have you tried running one as a VPN endpoint?
Another with a Celeron G1620 is running as an vsphere server. I really love the board.
On the other hand, even a sprinkling of installations is probably pretty valuable if it allows Google to derive more accurate ISP performance numbers. There's got to be some benefit in having thousands of devices directly connected to the ISP's network on the consumer end 24/7.
[0] http://www.amazon.com/gp/product/B005KG44V0?psc=1&redirect=t...
It does seem to have one USB port, though I'm not sure what it's for.
https://on.google.com/hub/#specs
It's my single biggest recommendation when people complain about the speed of the Internet in their home. More often than not, they're underspending on their networking equipment, not their ISP package.
Often, their networking equipment is part of the ISP package because the ISP wants to provide a seamless experience, and doesn't even provide easy access to the information necessary to use your own equipment.
Using your own router rather than the ISPs preconfigured router requires more than just plugging a personally-purchased router into their line (even in the simple case where you aren't using the ISPs bundled cable/phone/internet service that's all delivered through a common system.)
At least in the simple case, someone with modest technical skill can figure out how to get that information relatively easily, but most ISP customers aren't even modestly technical.
So what are you talking about? Purchasing your own cable modem? Or something else that I've never had to do?
Now, the ISPs that are using DSL (e.g., AT&T U-Verse which uses VDSL2 where it isn't FTTP) seem (from what I've experienced and seen with other people that have DSL-based service) to deploy all-in-one modem/routers (and, when they have services like TV and Phone as well as basic internet, to run them through the supplied all-in-one box, as well.)
> There's no requirement for DSL to bundle the modem and router.
Sure, its not a theoretical requirement, just practically what ISPs seem to do now.
An interesting variant on all of this is FiOS, which is often brought into the house on coax which plugs into their router. It's possible to buy something that will bridge the coax to ethernet, but they're pretty hard to find.
I think this might be going away as they increase speeds. I originally had the coax, but then they replaced it with ethernet coming in to the house. I kept their router anyway because it actually seems to be pretty good.
The usual recommended approach I've seen is to keep the ISP-provided modem/router, plug your router into it, and make certain configuration changes on one or both of the devices (though, for the same ISP, I've seen conflicting information on the correct configuration changes, with clearly dated assumptions, but still possibly correct as far as it goes information, in the official information from the ISP.)
But my basic point is that for most home users, they aren't conceptualizing it as spending separately on "internet service package" and "network hardware", the network hardware is part of the internet service package.
If that doesn't work, you can also just connect the modem-router's ethernet port to a LAN port on your own router, and just disable DHCP/DNS on the router.
My experience is that the DSL modems only contain mid-range routing and wi-fi equipment (10/100 switches instead of Gigabit, 802.11 b/g/n that run with one antenna instead of six and only on 2.4 GHz band). It's adequate to get things "working" but if you pay for a faster VDSL plan (usually 25 or 50 Mbps packages) you can see improved Wi-Fi performance with the right third-party router and laptop/phone.
Once you start looking at the latency requirements to achieve 10GBps interconnects you're going to have issues with what you can reasonably expect from the $200 price range, but consumer networks have a long ways to go before that's going to even need to be thought about.
Before I read into the product specifications I was expecting an Echo competitor via Google Now tech.
But with the a computer embedded in this router along with a speaker and bluetooth and zigbee radios this could be customized to offer a lot of the features that the Echo offers (minus the voice control). I hope that doesn't cause any problems.
This takes some processing power which not all routers have a lot of.
This link covers it a bit more in depth: http://www.pcworld.com/article/2689995/quality-of-service-ex...
I've set up and hand rolled my own Linux-based routers in the past on cheap leftover hardware (first gen Celeron). Adding QoS introduced literally zero noticeable CPU overhead.
I'm not saying you're wrong but you're clearly overselling the computational cost of this feature.
You are right. It isn't that much CPU, but with modern connection speeds (>=100 mbit) you actually need a half way decent CPU. Your old Celeron was probably routing a < 10 mbps connection.
Yes, your router uses much CPU for NAT. That's because it is a complex process which requires mangling the packets and keeping lots of state in a big table.
QoS however is relatively cheap. It's merely looking at fixed offsets in the IP packets and deciding which of those (mangled) packets must go now and which can go later. It requires some extra memory for some extra packet-queues, and a lookup in one of iptables' smallest rulesets. That's about it.
Yes, I routed things at slower speeds, but you got to have a perspective on things: You must NAT all the traffic, but only need to QoS the outgoing traffic.
This means that if the downstream speeds have increased 100x recent years while upbound speeds only 10x (which is my fair assesment of the market), your NAT cost is now 110x higher, while the QoS cost is merely 10x.
And QoS was the cheap load. If your router is already buried in other work, I'm sure QoS will feel extra painful, but so would anything else. On a healthy load you probably wont see it in the load-graphs at all.
Also worth keeping in mind is that QoS as a feature has diminishing returns the higher upload-rates you have. If you have 128kbps upload, making sure the right packets get out first is excessively important. If you have Gigabit upload, not only will QoS be more expensive, but you are almost guaranteed to have no benefit from it what so ever.
So yeah, I stand by my argument. QoS is not a heavy feature, even though marketed as such.
Inspecting every packet as it runs through the network takes a lot of processing power. Many router manufacturers put just enough hardware in the box to handle fast connections, but not enough to handle complicated QOS rules on top of that.
I guess that creates issues with interference with the antenna functioning though?
More at: http://business.financialpost.com/fp-tech-desk/cio/google-in...
What do they mean by "OnHub has 4GB of storage space, so there’s plenty of room for auto-updates and the latest software features." ? Is software for a router that large ?
So the question becomes, why not?
https://support.google.com/onhub/answer/6257015
As far as I know, the Almond+ is the only other router with 802.15.4 support.
2. On top of that, the specs are good enough to stream multiple IP cams.
3. Shot in the dark, but with all the "six 2.4GHz antennas, six 5GHz antennas, and a congestion-sensing antenna" they're probably trying to fix the classical WiFi congestion and star-topology range issues that comes with WiFi IoT. Especially because it supports 128 devices in WiFi. 4. Rumors are, it already runs Brillo, their IoT OS they showcased at IO. 5. Its bridge mode support be something they can use to have "multiple BRs" as promised by Thread?
[4] http://nextmarket.co/blogs/smarthomeweekly/29309313-google-i... [5] https://support.google.com/onhub/answer/6275097?vid=1-635755...
With its "advanced and always-evolving security features", I'm wondering whether and what information Google would be gathering about my local network and my network traffic.
"You agree that Google may collect and use technical and related information, including but not limited to information about your computer and/or mobile device, operating system, peripherals, applications, connected devices, network traffic, and data use to facilitate the provision of the Software and Services, including support and other related services."
It goes on to state "The OnHub Privacy FAQ describes the categories of data collected and how you can use privacy settings to change what data is collected by the Services."
But I couldn't find the OnHub Privacy FAQ and https://support.google.com/onhub/search?q=privacy oddly says "Your search - privacy - did not match any answers in OnHub Help".
Unclear why they refer to it as "The OnHub Privacy FAQ" (implying proper noun) but then the page is called "OnHub, the Google On app and your privacy" and it's not structured as an FAQ. I can't find anything else it could refer to, though.
https://www.google.ca/webhp?sourceid=chrome-instant&ion=1&es...
According to https://support.google.com/onhub/answer/6279845
Here in the UK most providers seem to offer the router as a sign up bonus.
The disadvantage is that router is mostly cheap garbage, but it's effectively yours once your contract starts, no rental/need to give back
If you don't have an expensive product, do some tests with and without the router's QoS. You'd be better of with a raspberry Pi doing QoS in most cases.
Because $40 routers can do that.
No $40 routers that I know of do anything like that.
So it's conceivable that it would better your Internet.
However, seeing better performance 5 GHz 802.11 n/ac requires you to have both a good router and a good laptop -- my experience is that unless your users/family blindly uses Macbooks, they won't see any improvement from a $200 router because their 2-3 year old $500 Windows laptop only has a 802.11 b/g/n 2.4 GHz-only 1x1.
My Airport Extreme chooses channels automatically at boot time, then never again. This is fine for my house though, because my neighbors are reasonably far away.
At work, our Meraki AP has this kind of 3rd radio, and we get very good performance in a terrible RF environment, where lots of other equipment (from Apple and Ubiquiti) has failed.
Google is doing nothing special here.
Google's spec sheet [1] specifically has 3 radios:
http://i.imgur.com/x7CMktb.png
The first two radios do MIMO 3x3, and the third radio just has 1 antenna (since it isn't serving client traffic).
While this isn't completely new, it's definitely not common, especially for a consumer-grade AP/router.
[1]: https://on.google.com/hub/#specs
For me, at least, a 3rd radio will be on the list for my next AP, so I'll take a close look at this one, too.
The issue is just that AVM produces different versions of the Fritz!Box for every ISP, so some have more storage, or more radios than others.
So, it’s pretty much in every house in Germany.
If OnHub has to compete against this (for the consumer free) router, then it’ll have a hard time.
Charging $199 for a TP-Link device sounds... overpriced. If this had been built by ASUS, then maybe they'd have something.
> Google’s hardware partner for the initial OnHub is TP-LINK; it plans future devices with other manufacturers as well, including one from ASUS later this year.
http://business.financialpost.com/fp-tech-desk/cio/google-in...
Including Bluetooth, Weave, and Thread, along with up to 128 devices, gives me a rough sense of the scale of connected devices they anticipate in each home. Given that it also has a reasonable compute capabilities and cooling as one of the highlights, I think it's far to say that it could do a lot of processing on its own before reaching back home to the GOOG DCs.