I'm only a little sad about this. Linksys was my go-to for hackable wireless routers until I discovered Ubiquiti. Seriously, these guys make some awesome stuff.
$80 will get you a legal-maximum 1000 mW transmitting, fully OpenWRT-compatible router with 8M flash, 32M RAM, a PoE injector, outdoor enclosure, mounting bracket, and a standard replaceable RP-SMA omni antenna.
We used the Ubiquiti wireless access points at my last job. Aside from being awesome for the price, the management software was a cinch to install on Ubuntu and seemed quite cool (what little I saw of it). They really seem to know what they're doing.
How exactly is that not a router? It takes packets and decides which of several interfaces to transfer them out on. That's the very definition of a "router". It even does such exotic stuff as dynamic NAT, firewalling, and DHCP provisioning.
It is indeed a semantic battle. But as for semantics: sorry, that's just wrong. A switch connects the same network transparently. A device doesn't need to know that there is a switch in place to find, say, 8.8.8.8. It just ARPs for the destination address and sends the packet. If you are connected to a wifi router, you must send a packet destined for 8.8.8.8 to the router, because it is your gateway.
I think we agree, but looking back my sentences are ambiguous. I slashed a lot of content out and lost the transitions. I meant that is the definition of a switch (not a router) and that a router does the following things, not that a switch is a router.
Routers will traditionally have more than a 180MHz CPU to deal with all the duties a router must contend with, they also have more than one Ethernet port, you know for WAN and for Wired LAN. This will only work as a Router if every device on your network is wireless, either that or you have no internet.
You're using a marketing definition for router, apparently meaning a (presumably expensive) rack-mounted piece of IT equipment. But traditionally "router" has been a technical term, and your wifi hardware absolutely meets the requirements.
Is NAT not a routing protocol? It certainly seems so to me -- you dynamically change your packet destinations, error behavior, and even their contents based on dynamic state detected from the network.
But yes, you mean OSPF et. al. Yes, you can certainly run those on a wifi router (they don't from the factory, so reflash with dd-wrt or whatever). Machines much less powerful were doing dynamic routing using these protocols 20 years ago.
Have you tried Mikrotik Routerboard? I know they are cheap and wireless oriented too (but they use proprietary router sw) but I've never had the opportunity to try one.
http://www.mikrotik.com/http://routerboard.com/
I haven't personally used Mikrotik products, but I know a few people who have and they really like them. It's on my list of things to buy and try out when I have time & money
I did the same thing. my Asus + tomato firmware = sanity. I couldn't believe how easy it was I've had the thing for about 4 years and never have to reboot it or anything.
New Asus routers like the N66U used TomatoUSB as a base, creating their own custom version of it. Then there are custom versions of Asus's firmware, such as the AsusWRT Merlin builds.
I suspect that the Ethernet switch is not part of the SoC, so it's a separate component that is already price-segmented. A consumer wireless router with 8+ Ethernet ports would probably not have significantly better margins than the existing models, but might need another round of FCC certification, and might also cannibalize sales of enterprise routers.
Also, I think most consumers who have a need for more than 4 computers to connect to their router via Ethernet probably don't want their entire network to have such an expensive single point of failure. I've had Ethernet ports on wireless routers die before. I like my switch separate from my router/AP separate from my modem.
i swore by WRT54GL w/sd card mod for many many years (on dd-wrt, then tomato). jumped ship to an Asus RT-N16 recently cause i needed the wireless-N and gigabit switch. currently sporting the Toastman mod of tomato, Shibby mod is also good choice.
I agree with you, the latest products were quite limited.
My brother bought one of their ADSL routers: the wireless range was limited (a lot less than my draytek and i think less than the old dlink we had), some features were crippled (usb port cannot be used for printer sharing but only for disk sharing, the disk should be fat and not ntfs, the sharing feature was basic) and wifi was continuosly freezing also with latest firmware 'til we searched on google and find some wifi parameter tuning hints on a forum
> I've never been satisfied with any Belkin product I've owned.
I've got a Belkin N+ router and boy is it bad. Have to restart it every two days to get to the maximum download speeds. Wlan(s) just drop at some point during the day and have to restart the whole thing. Argh.
Googling just now it seems that you can push the dd-wrt firmware onto it though. But I hope this move to buy syslink will bring something more to the table.
Every Belkin product I ever bough had problems (KVM switches and routers), so I just stopped buying from them, even cables. Now I'm not going to buy from Belkinsys too.
It could be great news. Linksys has some pretty good quality control and build quality, which belkin is lacking in. However, Linksys has always had serious problems when it comes to shutting up and getting out of the way. From what i've seen, belkin understands better than most that consumer networking equipment should be silent and invisible - it doesn't need to install software packages or pop up messages in your notification tray.
You really can't judge linksys by the wrt54gl - that's a product from another era, long before they were bought by cisco.
> From what i've seen, belkin understands better than most that consumer networking equipment should be silent and invisible - it doesn't need to install software packages or pop up messages in your notification tray.
As the purchaser and support, either personally, or through family, roomates, co-workders, and one landlord, of no less than Fifteen Linksys WRT54G[L] in the last 10 years (with the obligatory OpenWRT Mods), and a network engineer/manager by occupation - I look forward to seeing if Belkin can come up with a <$200 router that reliably works for more than 3 years. I think I've seen about 50% failure rate in a 3 year period for those fifteen Linksys WRT54Gs, as compared to the dozen Cisco 2621s that I have reaching year ten w/ 0% mortality.
Yes - I realize that a $75 Consumer WAP is going to be less reliable than a $2000 commercial router, but I would hope, over time, someone would come up with a inexpensive, reliable, workhorse consumer router.
Now that Belkin is acquiring the Linksys name, maybe they'll deliver it.
While everybody is complaining about Apple's AirPort being too epensive, of my ~15 devices that I have deployed for myself and for others in the last 8 years, none have broken down.
Also in all this time, none have crashed and needed to be restarted.
I realize that a web interface for management would be nice. I realize that I pay a premium for the brand. But I also think that there's a value in not having to replace devices continuously or just having to explain people how to restart them when they crash. For me, this value is worth the premium.
I had a Dlink ADSL router (in the 100 Euro range) and it died after 2-3 years (i think 2004-2006). Then I bought a Draytek router (because I needed VPN features), it was pricey (250 Euro) but it has since lasted for years (I'm still using it after 6-7 years).
For SoHo I think an Alix or similar hw running Pfsense could be a good choice.
When I was a student a lot of the houses I or my friends lived in had belkin routers/wifi access points. I don't think I encountered a single one that worked reliably, so I wouldn't hold my breath if I was you.
An idea why my Tomato flashed router needs a weekly reboot? I can't tell if it overheats, or what, but it gets to a point where it just stops responding: dhcp, WIFI association, etc.
Cisco neutered/limited the specs of Linksys WRT devices when they purchased Linksys so the offering wouldn't poach on their higher-end offerings. For instance the WRT54GS line used to have 32mb ram and 8mb flash and it was lowered to 16mb/2mb: source: https://en.wikipedia.org/wiki/Linksys_WRT54G_series#WRT54GS under Cisco. There is huge potential for beefier hardware that can run open source linux or bsd in this area. Hopefully Belkin will see the market opportunity and start creating some affordable routers that are hackable and powerful running 802.11ac. They'd sell millions imho.
Sounds like a win for me, concentrating all the terrible products in one company. I have no idea how Belkin stays around, I assume its due to a 500% markup on cables or something.
Cisco was always a poor fit for Linksys as they were scared of cannibalizing their own high-margin products. This market is getting hammered by TP-Link which provides consistently good quality products at low prices and many are DD-WRTable. Why buy a mid-range Linksys when you can get a very powerful Draytek et al with extra WAN ports and 3G backup for the same price?
Speculating that since the terms are not disclosed, they are not material to Cisco's earnings. That suggests they sold it for a lot less than the $500M they bought it for[1].
I wonder why they couldn't figure out how to make it work. I blew it by not working on NetApp's low end box (the Storvault) I really should have, it would have helped me understand where the sticky problems in 'consumer' are.
The StoreVault's key problems: artificial limitations, awful 'consumer' AI, and lack of SAN experience in their target audience.
The StoreVault ran a modified version of ONTAP, but still generally had most of the ONTAP features. Unfortunately, these were hidden by the special StoreVault System Manager. Even if you were familiar with NetApp, it was a major struggle to perform your typical tasks. The System Manager was buggy, slow, and only had the most basic workflows available. If you manually connected in to FilerView, you received a nasty warning about how dangerous/unsupported it was. But from FilerView, you could actually manage your filer the right way.
Non-savvy users struggled with the software and were steered in the wrong direction. It was incredibly frustrating being limited by the System Manager.
Now, the low-end filers like the 2040 occupy the same space, and are "full" NetApps so you don't have to worry about an abrupt EOL notice.
I just went to see if I could find the "2040," but it seem the lowest-end I could find was the v3140. Am I missing something? Older model?
Anyway, when I went to see what these appliances' specs are and what they would sell for, it was obvious why they failed at the consumer market:
These high-touch sales models where one is required to talk to a salesperson, and there are a ton of configurability options will not work in the consumer market.
Exactly, the configuration options are daunting, and your average consumer doesn't understand a SAN architecture well-enough to make the right choices. It's sort of a catch-22, where any company looking at a low-end SAN also doesn't have the personnel to run any SAN effectively.
NetApp's FAS product line has 3 families: the 2xxx series, the 3xxx series, and the 6xxx series. Each of these families then also has sub-groupings based on release date/capabilities, like the 20xx series, 22xx series, etc. So the 2040 is their lowest-end filer they still sell (and that supports the latest OS version), while the 2240 is the newer build designed to replace the 2040 (better CPUs, RAM, etc).
The 2xxx series (2040 (older, but available), 2220, 2240) are all lower-end and have limited expansion capabilities, as well as including disks in the controller chassis. The 2240 is unique in that it can support a single 10Gb ethernet card or an 8Gb FC card (but not both at the same time), while the rest of the 2xxx family can't.
The 3xxx series (3140 (older), 3220, 3250) are mid-range filers with decent expansion options (SSD/Flash PCIe card, 10Gb ethernet, 8Gb fiber channel, more SAS ports, etc).
The 6xxx series (6040, 6080, with the 62xx series coming soon) are high-end filers with tons of expansion ports and built-in ports to handle larger workloads and higher total storage limits.
The v-series is designed to be a front-end to another SAN, giving you the NetApp suite of software/capabilities without rebuying all of that disk.
There's also the E-series, which is the Engenio tech they're selling after acquiring them a year or two ago. I haven't worked with that, but it's the very high-end filer.
Cisco is built bottom-to-top around enterprise sales. Linksys is a consumer company. Combining those two after-the-fact requires a lot of willpower that few companies successfully manage.
So if Belkin hasn't ever made good hardware, and Linksys hasn't made good hardware in a few years—what company is making solid consumer networking hardware at a decent price?
I know next to nothing about networking hardware, so fuck me, but...
I always had to power-cycle my routers, no matter the brand, for years. Some once a week, one every other day. Then I bought a Time Capsule from Apple. It seemed kinda pricey, but I wanted the convenience of the backups.
This was in 2008. Five years later, that thing still works perfectly. Never had to power cycle it. Entirely reliable. So Apple will keep getting my money for consumer networking gear.
It's good that you haven't had any problem with your Time Capsule, especially overheating. Due to poor engineering and industrial design, there are a lot of people that have had issues with that and had to physically modify the case [1].
In general Apple has a solid history of various problems (I'd speculate mostly due to design decisions taking over engineering concerns) with their hardware - "antennagate", Mac Mini overheating, dust inside DVD drive slots, Time Capsule and Apple TV overheating [2].
I guess there are more, but these are the problems people around me have actually stumbled upon.
Not exactly what I would call a manufacturer of durable hardware.
Asus. I've had a rock solid WL-500Gp and 2 WL-520gU's. They're running DD-WRT, but the stock firmware would probably be ok since the hardware's been excellent. I'm considering the RT-N66U next[1] which is a little pricey but ranks well on SmallNetBuilder throughput charts[2] and supports a custom firmware build that uses the native optimized drivers[3] (or DD-WRT/OpenWRT, etc.).
Also, any Linksys router with an external antenna has been OK, but those are rare now. Netgear and D-Link have been hit or miss for me.
//edit: A couple folks have said Apple Airport. +1 from me as well; I've used one without any incident for a couple years connecting primarily Windows/Android machines.
Basically nobody, its pretty hit or miss at this point. The only device which I'd say is solidly worth buying is the Apple Airport, however its pretty low on features.
I see a fair amount of these sorts of devices and I can tell you most of them make me want to stomp on them and swear at the people who buy them. They're so terrible.
I know that some people love them, but as someone who does network engineering for money, people need to get real.
I use a home-built router and a TP-link WiFi access point, works great! Building your own router is as simple as having a Linux box with two ethernet plugs and installing a DHCP server, and some iptables tweaking for NAT.
Seems like overkill power wise though, right? I wouldn't mind repurposing an old box for that either but even something with a 250w lightweight desktop PSU is gonna out-juice a little embedded device.
That depends. There has been low-power CPUs along the history[1], if you manage to find a box with one of those then I wouldn't be surprised if you get way below 50W. Of course that probably is still significantly more than what an embedded solution would consume, but on the other hand you should get better performance and more flexibility. You could use the same box for eg NAS.
[1] Intel ARK finds 108 =<35W desktop CPUs, from the original Pentiums to modern day i5 Ivy Bridges.
Linux runs on Linksys routers (dd-wrt, for example), so Linux != lots of power :) There are some cool boards you can get pretty cheap with 10/100 ethernet, which is probably all you need for an internet conneciton (I have 60 up/down at home so it's enough for me).
I've got a TP-Link router I've been using for the past year and a bit. Works great, no stability issues, and has all the hacky config options I need right of the box.
I had a terrible Zyxel router was really unreliable. Replaced it with a Billion 7404VNPX and haven't looked back. They're aimed more at home/small office than just being consumer, so they have better configuration, ipv6, etc. (but are a little ugly, but who cares?). Still pretty cheap.
My Netgear WNR3500L has been nothing but rock solid since I bought it. The router is advertised to be compatible with Tomato, DD-WRT, and Open-WRT. I think I have had to reboot the router a grand total of one time, and it was completely my fault.
Buffalo makes decent hardware (don't count on the software though), so does TP-link. But in any case the OEM firmware is most probably horrible, so most important factor imho is OpenWRT (or some other distro) support.
hardware is cheap and commodity. your best bet is looking through dd-wrt and finding what hardware offers the best specs and easy to install this firmware
In one of those "I have no way of knowing that it's true, but it's still almost certainly true" predictions: For the last 10 years:
- Cisco upper management has been yelling at the Linksys guys to add features and do things "the Cisco way".
- Linksys engineering has been subtly hinting back to the Cisco guys if they added those features and did things the Cisco way, a low end router would cost $1000.
And most of the best Linksys guys acquired either work for another division within Cisco or a different company altogether, because they saw the writing on the wall a long time ago.
does anyone know why Cisco sold Linksys? As far as I recall, Cisco wanted to enter the consumer market (as opposed to their enterprise presence).
Did they fail?
I like the fact that Cisco is not afraid of letting go stuff they acquired. Last time I heard was when they acquired tribe.net and later put it up for sale? correct me if I am wrong. And then there was the acquisition of Flip camera, that flopped (no pun intended).
I wouldn't call Meraki a consumer product though. Their cheapest AP is $400 and you have to buy per/AP license at $150/year. Not something I'd put in my parent's house.
I recall reading somewhere that the Linksys devision was more or less going downhill fast, due largely to cable/dsl modem vendors including wireless in their models, and operators simply using them for customers. Can't remember where I read that, or if it was even true, so consider it a rumor I guess.
They made really bad bets on consumer products. For Linksys they bought it to get into the entry level / small business market as well. Another bad bet. The competition has been stepping it up to the point where these products overlap with Cisco's existing platforms. Spending the money to develop a separate hardware/software platform doesn't make a lot of sense. They can just downsize/license option their existing platforms into that space and save themselves a ton of money.
Cisco focusing more and more on software and services. Interesting litmus test will be their hard VC and telepresence endpoints. Enterprise market and Cisco a leader...however Cisco knows future is soft clients, UC, video-enabled apps, services...how quick do they leap?
http://www.ubnt.com/
$80 will get you a legal-maximum 1000 mW transmitting, fully OpenWRT-compatible router with 8M flash, 32M RAM, a PoE injector, outdoor enclosure, mounting bracket, and a standard replaceable RP-SMA omni antenna.
http://www.ubnt.com/picostation