Cisco Linksys E4200V2 Maximum Performance Dual-Band N N900 Router

[Shikami]

But my point still holds that when an ISP provisions X amount of bandwidth, there is nothing that a router can do to get more bandwidth from that service.

Absolutely, correct. However, how well the router handles the packets in transit with multiple concurrent users is imperative. If the amount of allocated bandwidth is scant and asynchronous, with multiple users, QoS will be absolute; obviously due to low bandwidth. No amount of PPS and packet control can really help with the issue of low bandwidth. Latency will increase to their point of origins, but should/can be somewhat reduced due to QoS, even though QoS can induce latency due to packet buffering. Still, a more capable router will show some benefit in this case with the handling of packet transits better and faster, albeit minimal.

When the bandwidth allocation is higher, then a more powerful router can really shine with multiple concurrent users. But to note the main resolve for little bandwidth, is more bandwidth of course. In this case, again though, a more capable router will prevent increase of latency and throughput.

Basically, when my roommate, their significant other, and friends come over and uses my connection to the Internet, and I am gaming I do not want to even know that it is being used by someone else. To me working with the amount of allocated bandwidth is important, even if it is at your home.

 

[rhombus]

Not to create contention

But to note the main resolve for little bandwidth, is more bandwidth of course. In this case, again though, a more capable router will prevent increase of latency

Shikami, intended or otherwise, you make me laugh

 

[Unregistered]

Hi
One thing I notice no one taking into account or talking about, internal traffic. Am I wrong to assume this router would also be slower at internal traffic ?

Wouldn't it speed barrier slow multiple conection transfers on the internall side?

And last I am really eager to upgrade after trying the netgear, having wired speeds on wireless is so much easer than routing cables all the time.

So my last qwestion is can its speed be fixed threw firmware?

Thanks

Hmmm, i believe the test results where around 200mbs +- thats not anywhere near gigabit, i guess that means files from computer to computer on LAN side would be slower.
Anyone confirm this ?

 

[Shikami]

Hmmm, i believe the test results where around 200mbs +- thats not anywhere near gigabit, i guess that means files from computer to computer on LAN side would be slower.
Anyone confirm this ?

Passing layer 2 frames to another MAC is nothing much for the layer 2 chip used. You are more likely to have slow transfers due to the medias and the subsystems of the hosts used. You cannot use the results of the speed tests to indicate what you would get for Ethernet usage, LAN to LAN. Because the tests for such have not been done.

 

[rhombus]

Translation: don't worry, LAN to LAN can operate at full gigabit speeds.

 

[Unregistered]

Passing layer 2 frames to another MAC is nothing much for the layer 2 chip used. You are more likely to have slow transfers due to the medias and the subsystems of the hosts used. You cannot use the results of the speed tests to indicate what you would get for Ethernet usage, LAN to LAN. Because the tests for such have not been done.

What worries me is I have a few full HD ip cameras when set to jpg 2048x1536 they use a lot of bandwidth 2217.98kb/s or more each camera, on top of that a server and a TiVo with steaming video and sling player. Now my cable internet speeds are 120mbps dowload and 15mbps upload. I have treid every new top router out and still fall back to a older dlink for stability and speed and strangely enough it was testest as fast and is still top speeds on charts yet all the annoyances like the fan and broken switch I do not have. I treid the new netgear 4500 and it was fast, did I say fast yet. Lol
Though it's stability was something to be desired and would slow to crawl with just connecting a hp office jet printer or my Linux server and even if nothing conected would not conect to Internet after router reboot, for some reason Turning off its wireless then it would conect, this is an ishue since I loose power a lot and servers must be running 24/7 with out ishue. my dlink link is aging and really not up to its task even though I can rely on it to be up and running 24/7

So I am pondering this Cisco for its good build, cpu and from what I read on its stability. Just tired of trying routers only to have to place them in storage.

So any one know if the Cisco speed problem can be corrected with firmware or is it more likely a hardware design flaw? Thanks

 

[Shikami]

Translation: don't worry, LAN to LAN can operate at full gigabit speeds.

Actually I said: The chair is against the wall. The chair is against the wall. John has a long mustache. John has a long mustache.

 

[sfx2000]

Getting the thread back on track - bandwidth across the LAN should be similar between the V1 and the V2 devices...

I think Shikami's concern, and it is a valid one, is the performance delta on the WAN side of the router between the Broadcom and Marvell solutions as implemented in the E4200, not that the V2 is bad, just that the observed results are so different...

My guess is the new software stack that Cisco-Linksys is implementing... the hardware itself is more than capable on a WiFi, host AP, and switching level.

Looks like there is a lot of headroom for improvement considering some of the results (the SMB for example). It's a matter of Cisco-Linksys and if they want to invest time/effort at making this particular product fly...

 

[rhombus]

I haven't seen any valid concerns.

Routing speed with all default features engaged is still many times more than a home internet connection, USB file transfer speeds are up 3-4x, and simultaneous connections are enough to sustain 40 computers running BitTorrent on its default settings.

Anyone claiming routing performance is hobbled by Cisco software/firmware and that this may therefore indicate something more sinister lurking underneath, needs to first identify the Marvell SoC used and then compare this to routers using the same chip. It is a nonsense to compare with the previous Broadcom solution or to expect a direct correlation with CPU speed alone.

The only other claim I have seen is a "top speed" argument. It goes like this: the speed limit may be 70mph and the top speed of the car may be 120mph, but a different car with a top speed of 200mph is a more capable car, and will therefore always handle corners better even at 70mph. Not only is the argument flawed, but it may well result in conclusions that are upside-down.

The idea that configuring QoS rules (+ any default, auto QoS) and then sending multiple traffic classes will slow down traffic/increase latency to below that of a home internet upload is pretty far-fetched, and that's even if we allow the flawed top speed argument. And things that probably will make a real-world difference, like VPN tunneling, need to be benchmarked separately, but even here I wouldn't be surprised if V2 wins out because this is usually limited by CPU speed.

If anyone has benchmarks or tests that show real-world performance hits, let's see them. Until then, I would say keep the conjecture and hysteria down.

 

[Shikami]

Well, one point that cannot be refuted, is if a consumer is purchasing and expecting a Ferrari. He better get a Ferrari, and not a Pinto. Basically, version 2's cost does not equal with its price to performance ratio. Version 1 does, without a doubt. Why pay the price premium for a worse or the worst performance?

 

[Unregistered]

I haven't seen any valid concerns.


The only other claim I have seen is a "top speed" argument. It goes like this: the speed limit may be 70mph and the top speed of the car may be 120mph, but a different car with a top speed of 200mph is a more capable car, and will therefore always handle corners better even at 70mph. Not only is the argument flawed, but it may well result in conclusions that are upside-.

That is upside down , you can't compare cars to routers. I know cars and any car that has a speed of 200 mph will handle better, it will have the best of the best that money can buy. This is a fact. Any one that disagrees knows nothing about cars

Back to subject
Point is well taken. a router that has top speeds may not be as stable as a lesser router.
My Personal experience I have had a few of the top rated routers with the fastest benchmark speeds slow to crawl or just lock up from what seemed to be a compatibility problem with linux servers or a wireless printer and the router or should I say the routers firmware?

This never happend with my hardware and any other of the tens of routers I have used over the years,
New technology, got to love it,maybe they work things out over time

 

[sfx2000]

Well, one point that cannot be refuted, is if a consumer is purchasing and expecting a Ferrari. He better get a Ferrari, and not a Pinto. Basically, version 2's cost does not equal with its price to performance ratio. Version 1 does, without a doubt. Why pay the price premium for a worse or the worst performance?

The HW is more than capable of matching the observed performance delta.

Other platforms use the same core logics (processor/switch/radios) and get very good performance. I was a bit surprised to see the delta's between the v1 and v2 (not that the v2 is bad, but that the delta was even present). Wasn't until going back for a re-read did I see that Tim noted that Linksys made a point of the new SW stack.

 

[Shikami]

What worries me is I have a few full HD ip cameras when set to jpg 2048x1536 they use a lot of bandwidth 2217.98kb/s or more each camera, on top of that a server and a TiVo with steaming video and sling player. Now my cable internet speeds are 120mbps dowload and 15mbps upload.

I understand completely the concerns of your network, although some here argue that it is not important . Curious what routers have you tired, and what basic issue did you have with each of them? You or others may discover that there is an issue with your network.

Just a note about some of the issues you mention. The routers can have a common function/bug that if I/O is too high it can reset the router. This is a common issue with many routers and is usually resolved in due time. There can be many driver issues that are resolved slowly due to the very nature of Linux. What can make it great is the very fact that it is Linux. The community can take the sources and make it better, usually fixing the bugs, even increasing the performance and adding new features to the router. The router I use, the RT-N56U, had a few nasty bugs with the HW NAT acceleration that Asus could not fix and deemed them impossible. However, 3rd party firmware fixed it easily, and so far has made the router better.

I am thinking that you may actually want to try making a firewall/router using pfSense (http://www.pfsense.org/). It can be costly, but has more features, and abilities with the excellent stability of Unix. Finding the right micro board can be arduous, IMO. This is because I am particular about my networks and the hardware that is used. Some take a simple Atom processor based micro board and use it as a system. I like a bit more processing power, but not too much that it is another power hungry computer.

 

[rhombus]

If anyone can devise a straight-up test, of their design and choice, to actually demonstrate and measure any performance inadequacies that might be encountered when connecting the WAN to a typical home internet connection, then I will happily get my hands on a E4200V1 and V2, carry it out to the letter and genuinely post the results.

I can do the same for an Asus RT-N56 running stock firmware with its two very different WAN-LAN performances (hardware NAT enabled/disabled), and see its effect.

I have IxChariot for several Windows endpoints and plenty gear to work with. The test should be controlled, repeatable, simulate the real-world and be written up as a short, clear, objective test protocol.

I'm sure those posting will be eager to prove their point with this simple enough task. And not just that, if a success, this test can then be recommended for Smallnetbuilder too.

Edit: Offer withdrawn, lol. I was serious about doing it - someone else might like to carry it out though, I think it would be good to do.

 

[sfx2000]

If anyone can devise a straight-up test, of their design and choice, to actually demonstrate and measure any performance inadequacies that might be encountered when connecting the WAN to a typical home internet connection, then I will happily get my hands on a E4200V1 and V2, carry it out to the letter and genuinely post the results.

I can do the same for an Asus RT-N56 running stock firmware with its two very different WAN-LAN performances (hardware NAT enabled/disabled), and see its effect.

I have IxChariot for several Windows endpoints and plenty gear to work with. The test should be controlled, repeatable, simulate the real-world and be written up as a short, clear, objective test protocol.

I'm sure those posting will be eager to prove their point with this simple enough task. And not just that, if a success, this test can then be recommended for Smallnetbuilder too.

Edit: Offer withdrawn, lol. I was serious about doing it - someone else might like to carry it out though, I think it would be good to do.

Funny comment, but seriously, something to consider is a collaborative test plan - between Tim, Stevech, Shikami, and yourself - the collective experience could be a major plus to this site - I'd be happy to help out with test case generation/review as well (having had some experience at this).

Thoughts?

 

[Unregistered]

I understand completely the concerns of your network, although some here argue that it is not important . Curious what routers have you tired, and what basic issue did you have with each of them? You or others may discover that there is an issue with your network.

Just a note about some of the issues you mention. The routers can have a common function/bug that if I/O is too high it can reset the router. This is a common issue with many routers and is usually resolved in due time. There can be many driver issues that are resolved slowly due to the very nature of Linux. What can make it great is the very fact that it is Linux. The community can take the sources and make it better, usually fixing the bugs, even increasing the performance and adding new features to the router. The router I use, the RT-N56U, had a few nasty bugs with the HW NAT acceleration that Asus could not fix and deemed them impossible. However, 3rd party firmware fixed it easily, and so far has made the router better.


I am thinking that you may actually want to try making a firewall/router using pfSense (http://www.pfsense.org/). It can be costly, but has more features, and abilities with the excellent stability of Unix. Finding the right micro board can be arduous, IMO. This is because I am particular about my networks and the hardware that is used. Some take a simple Atom processor based micro board and use it as a system. I like a bit more processing power, but not too much that it is another power hungry computer.

Funny you say that, the one problem I have is router will reset itself ( compleat reset back to defoults ) and this happen with every router at some point

As far as the routers I treid well I take my 10 grand back please, of course the top 15 smallnetbuilder list, i try any new model that is faster then last years. haha netbuilder dont allways review all models and i allways want to send them to netbuilder for review. Treid Cisco, netgear, Asus n56u was the biggest flop for me, just froze up most of the time and wireless had no range , the netgear 4500 had longest rang for me and was getting the full 120mbps over wireless to 2 stream hawking USB receivers. Strange it slowed to crawl when coected wirelessly to my hp office jet printer, yet my linux server didnt bother it. it was pretty nice and fast just had its own firmware issues that everyone is aware of including Netgear

I been stuck with and allways fall back to strangly enough dir-685, it handles everything including 2 acces points a 2.4 and 5 MHz . I get decent speeds and stability, i do have to reboot the router every now and then and just about once a year it decides to reset to factory defoults
I have to say the netgear4500 was a lot faster , did I say faster yet oh by the way did a mention the netgear4500 is super fast, lol


Now let me say we have tested each device on its own to narrow down the issue and yes it's the old Linux server on the most part and for some reason also an hp office jet wireless conection, there's nothing I can or will change about that till there s a suitable replacement I expect one day
I using a older version of Linux fed11 for server and can not update it do to the software I using.

Yes I have treid untangle an it's faster and no real ishues to speak, also handled my Linux server like it was nothing,
though We have 3 steam wireless N accses pionts have ishue with caped 60mbps wireless speeds on 2.4mhz and untangle, dont understand cuase when set to 5mhz full 120mbps speeds

You say pfSense is better ?
we using the newest Xeon 12 thread procesers for hardware with the newest 4 port intel pci-e servers cards

 

[Unregistered]

The HW is more than capable of matching the observed performance delta.

Other platforms use the same core logics (processor/switch/radios) and get very good performance. I was a bit surprised to see the delta's between the v1 and v2 (not that the v2 is bad, but that the delta was even present). Wasn't until going back for a re-read did I see that Tim noted that Linksys made a point of the new SW stack.

Ok then i agrre if the hardware is capable, there is still the simple qwestion why did it bench so slow? Simple question any one ?

Why is it is a less notable brand router had had the same benchmarks it would be considered bad ? But not linksys

Are we letting linksys pass a test with good gugment Just because of it name brand ?
If the same router was egenius or Roswell wouldn't it be a fail ? And nobody woul buy it
Shuoldnt a Ferrari be a Ferrari ? And if it's a lemon shuoldnt it be considered. Lemon

Why doesn't linksys just fix the firmware ? what if they don't or can't ?

What if a new hardware revision is needed ?

Oh well If the new dlink yet to be released doesn't make the grade guess I be using untangle or something like that

 

[Unregistered]

i need to clarify something i am not saying the linksys is a lemon, linksys used very good hardware

i am in hopeing that there just may have been a compatibility isue or firmware issue during the smallnetbuolder testing or maybe a defect in the router . lol that would explain everything as the asus n56r was a flop for me with its very limited wireless range and its freezing up and slow gui and it was not faster then the dlink i using in fact slower wired or wirelessly and it 5mhz was slower with not even half the range of my ap

i contemplating testing the e4200v2 myself i just don't have the time anymore to play with routers like i used to. From the user reviews the linksys sound pretty darn stable as i would expect so from its processor

 

[thiggins]

There is no "defect" in the router and Cisco has not questioned the results. The lower throughput is simply a design decision. The high routing throughput isn't needed for many/most consumers. Perhaps they are reserving the CPU power for future features.

 

[tipstir]

About time they give you 1GHz NPU and enough RAM for it. Faster it is the better though-put and the demands of 1080p streaming that I do now the E4200 is able to keep up with Amazon Prime Videos HD, Hulu Plus and Netflix. I have more of these 8x SONY SMP-N100 and 1x ROKU2 just ordered to see how well that does over 802.11N @ 40MHz.

I would like to see how well the new E4200V2 vs E4200V1 does. I might go get one and see for myself. But right now the E4200V1 with the 2x RE1000 work very well. E4200V1 can't cover my second home here. I don't see why it can just blanket the house. I still need to get additional RE1000 which I did. I would like to stream outside in my Patio with SONY Network Media Player or this new ROKU2 @ 1080p.

Seeing the review here ont he E4200V2 seems to be lower in WiFi than the E4200V1. How about MSC is has high as the E4200V1 though. I have one AMD Quad Core max out in RAM with PCI-E using 256KB Packet Buffer and set the TCP.sys set 200 MSC instead of the 10 default and it really screams quickly.

Update: Android App called Cisco Connect Express works fine on my E4200V1. I have that installed on all 4x Android tablets including 10.1" Thrive. Also my Android 4G Smartphone too with 802.11N @ 40Mhz. Of course I could use the Google Chrome Browser to access the admin web GUI for the router too.