dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
4856
share rss forum feed

Doeboye

join:2006-11-07
Canada
reply to EdT

Re: affordable router to handle 150/10?

said by EdT:

I just finished chatting with some friends oversea about their internet connections. I think I flipped on what they are paying and getting.

Hong-Kong and China = 100Mbps/100Mbps cable internet $23cad !!!

Gave him the link to SpeedGuide.net thinking he was mistaken, dam was I ever mistaken ! ...LoL

Did you ask them what kind of router they were using?


EdT

join:2009-06-12
Saint-Laurent, QC

1 edit
I did, he didn't know, he wasn't the techie type, but he said it was a cable modem and not DSL. I will ask him again the next time he is online.
»www.speedtest.net/result/2620356799.png


TwiztedZero
Nine Zero Burp Nine Six
Premium
join:2011-03-31
Toronto, ON
kudos:5
reply to morisato
said by morisato:

Small net Builder places that at 144 MBPS thruput is the main issue i have with it, So i spent the $89.99 for a WNDR 3700 from netgear got the v3 and installed my ddwrt and did a good old wifi Spectrum anaylsis my strength is damned good now that i have it looks like i killed some other networks trying to run on channel 1 p:)

+1 LOL
I still use my original WNDR3700v1, but am considering a new AC unit for my next buy.
--
----|- From the mind located in the shadows of infinity -|----
Nine.Zero.Burp.Nine.Six
Twitter = Twizted Zero
Chat = irc.teksavvy.ca

mlord

join:2006-11-05
Nepean, ON
kudos:13
Reviews:
·Start Communicat..
reply to technocar2
said by technocar2:

With my torrent server completely saturating my uplink I can still do a speed test and get full speed out my second VPN (private internet access); this is QoS in action at deprioritizing my torrent server.

Speed tests for PIA are more than a bit misleading: PIA uses compression on the VPN pipe, and the speedtest.net test data compresses rather nicely, resulting in somewhat unrealistic reported numbers. It's quite possible that the second VPN was only getting half the bandwidth there.

technocar2

join:2009-05-29
Brampton, ON
kudos:2

1 edit
said by mlord:

Speed tests for PIA are more than a bit misleading: PIA uses compression on the VPN pipe, and the speedtest.net test data compresses rather nicely, resulting in somewhat unrealistic reported numbers. It's quite possible that the second VPN was only getting half the bandwidth there.

I know what you are talking about; basically you are saying I only downloading half the "data" but its compressed; and once its uncompressed it doubles in size thus showing inflated speeds.

I have heard that argument a million times on these forum, and I'm not buying it because the pfsense WAN graph shows the real speed and it shows ~150mbps on the WAN interface graphs when my VPN connection is saturated on downlink so if it was compressed as much as you say it is then it would show that much less on the WAN interface graph and much higher on the VPN interface graphs. But the graphs are similar regardless of what you think (~150mbps on WAN interface graph and ~140mbps on the VPN interface graph). Until these numbers start being wacky; I'm not going to believe your argument.

Mate if what you say were the case then the WAN interface should show 70mbps and VPN interfaces should show 140mbps but that is not the case.

s0dhi

join:2011-08-02
Brampton, ON
Reviews:
·TekSavvy DSL
reply to Optional
I've been using pfsense running on an ALIX2D3 for firewall/routing and a WRT400N running DD-WRT dumbed down as a WiFi Access Point.

The setup has been rock solid on my connections thus far, but the ALIX would be a little light for the 150/10.

Supermicro has a few dual core ATOM boxes that would work great and have gigabit Intel NICs. Those running pfsense would be rock solid.

technocar2

join:2009-05-29
Brampton, ON
kudos:2
said by s0dhi:

I've been using pfsense running on an ALIX2D3 for firewall/routing and a WRT400N running DD-WRT dumbed down as a WiFi Access Point.

I have a similar setup but with different hardware.

pfsense on a core 2 extreme machine QX9650 for firewall, routing, VPN, captive portal etc. (honestly I didn't have any other use for this beast of a CPU)
Wifi AP on a dumbed down WNDR3700 running DD-WRT with a captive portal for wifi only users on pfsense (it took me a while to build a decent captive portal but its perfect now)

And like you said, its absolutely rock solid.

technocar2

join:2009-05-29
Brampton, ON
kudos:2
reply to EdT
said by EdT:

I did, he didn't know, he wasn't the techie type, but he said it was a cable modem and not DSL. I will ask him again the next time he is online.
»www.speedtest.net/result/2620356799.png

Correct me if I'm wrong but that speed test's ping is too low and upload is too high, that can only mean its FTTH.
He probably thinks its docsis but he actually has FTTH otherwise his upload won't be that high.

morisato

join:2008-03-16
Oshawa, ON
Reviews:
·TekSavvy Cable
·TekSavvy DSL
·ELECTRONICBOX
I am pretty sure it can handle qos and 150/10 i am running ddwrt on it and it seems to be okay so far, also i don;t torrent, so i don;t need it to handle thousands of micro connections like you might. and i had a pfsense box no thanks, if i go that route though i would pick up a raspberry pi.
--
Every time Someone leaves Sympatico an Angel gets its wings.

s0dhi

join:2011-08-02
Brampton, ON
Reviews:
·TekSavvy DSL

1 edit
said by morisato:

I am pretty sure it can handle qos and 150/10 i am running ddwrt on it and it seems to be okay so far, also i don;t torrent, so i don;t need it to handle thousands of micro connections like you might. and i had a pfsense box no thanks, if i go that route though i would pick up a raspberry pi.

Raspberry PI only has a single NIC and IIRC, the NIC is 10/100.

I don't quite understand the statement "i had a pfsense box no thanks". What did you not like about it?

EDIT: Also the RaspberryPi NIC performance is severely limited by being connected to the USB bus.

technocar2

join:2009-05-29
Brampton, ON
kudos:2
reply to morisato
said by morisato:

I am pretty sure it can handle qos and 150/10 i am running ddwrt on it and it seems to be okay so far, also i don;t torrent, so i don;t need it to handle thousands of micro connections like you might. and i had a pfsense box no thanks, if i go that route though i would pick up a raspberry pi.

The way dd-wrt works is if you even just turn on QoS it will dip your speed to whatever your CPU can support regardless of how many connections you have or how many policies you have, max I got was ~100mbps with QoS with just one computer and zero policy but the average was about 80mbps with 4 computers and 10 policies, if you don't believe me then don't. You can try it for yourself and then I'll say I told you so.

technocar2

join:2009-05-29
Brampton, ON
kudos:2

1 recommendation

reply to s0dhi
said by s0dhi:

I don't quite understand the statement "i had a pfsense box no thanks". What did you not like about it?

I think the complicatedness and the learning curve turns people away from pfsense. But once you get past it you start seeing the beauty that pfsense is. Its not for everyone I suppose.

s0dhi

join:2011-08-02
Brampton, ON
Reviews:
·TekSavvy DSL
said by technocar2:

said by s0dhi:

I don't quite understand the statement "i had a pfsense box no thanks". What did you not like about it?

I think the complicatedness and the learning curve turns people away from pfsense. But once you get past it you start seeing the beauty that pfsense is. Its not for everyone I suppose.

You do have a point there. I can't remember the early days of my cut-over from Tomato/DD-WRT to pfsense, but having stuck with it, I'm content (especially with the VPN and DNS poisoning).


TypeS

join:2012-12-17
London, ON
kudos:1
Reviews:
·TekSavvy Cable
reply to Optional
Hmm interesting, my venerable WRT600N is still viable 6 years later SMB Linksys WRT600N Review.

I'll probably be picked up the RT-N66u or RT-AC66u soon anyway and relegate the WRT600N to being secondary WAP for the other end of the house that gets low signals. Even with 170+Mbps WAN to LAN, it's used as all in one in my network right now, so I don't expect to be able to use it for 150/10.

Doeboye

join:2006-11-07
Canada
reply to Optional
Just an FYI,

NCIX has dropped the price of the TP-Link WDR4300 by another 5$ (To $64.99), AND have included free ground shipping.

Also, they extended the deal until April 10th.

»ncix.com/products/?sku=74472&vpn···oid=1277

[EDIT] The WDR3600, which also supports WAN speeds above 150mbps, is also on sale:

»ncix.com/products/?sku=77455&vpn···oid=1277

$49.99 and free shipping

Doeboye

join:2006-11-07
Canada
Another deal!

Staples has the more expensive (but better) Asus RT-N66U on sale for $149.93, and if you have a staples $15 off coupon, you can get it applied as well. I pulled the trigger on one and got it for $135 + tax.

»www.staples.ca/en/ASUS-RT-N66U-D···_1_20001

That said, it seems like I got the last online one, but I checked store stock, and it appears to be available in-store (in ottawa at least).

Though not inexpensive, certainly a pretty good price for this guy.

Also, the equally fast RT-N56 is also available, and with the coupon, you can get it for $95. Only reason I didn't go for this one instead is because it is not wall-mountable.

»www.staples.ca/en/ASUS-RT-N56U-S···_1_20001

You need to be signed up to their newsletter to get a coupon code mailed to you, or, you can go to redflagdeals.com and check out the forums. People post their unused coupon codes there all the time...


RMerlin

join:2009-10-09
Montreal, QC
reply to Guspaz
said by Guspaz:

The basic results of his test are that you need something newer than the RT-N16 to drive the new 150 meg service. It was a 2.4 GHz device anyhow, so a newer router can upgrade your wifi too.

It's not the CPU that matters so much; the RT-N16 and the N66U have virtually the same CPU/SoC (same CPU core, 25% higher clockspeed in N66U), yet the N66U can route like 8x more throughput. The CPU isn't doing the routing anyhow, it's offloaded to dedicated hardware.

Unlike their Ralink-based devices (like the RT-N56U), there isn't that much HW offloading on the RT-N66U. The only performance improvements come from when you have the so-called "HW acceleration" enabled, which does nothing more but load the CTF (Cut-Through Forward) kernel module, which essentially takes some shortcut through the network stack. That module only works if you don't use IPv6 (since FW 354), QoS or URL filtering. I must admit tho that it's hard to be 100% sure as to what this module does since it's closed-sourced. I think the RT-N16 also has that same module, provided you use a recent firmware (both routers now share the same code base).

Those extra 150 MHz make a big difference when it comes to packet routing. You could probably confirm it by taking an RT-N66U with an up-to-date CFE, and downclocking it at 450 MHz.


Guspaz
Guspaz
Premium,MVM
join:2001-11-05
Montreal, QC
kudos:23
So you're saying that 125% the clockspeed on a virtually identical CPU can produce 800% the routing throughput? That's a pretty extreme claim.

We're talking about benchmarks that show the N16 doing ~100ish and the N66U doing ~800ish here, and it's the same CPU architecture (literally the same core at a different clockspeed). I'm going to argue that it doesn't make any sense at all that the N66U is doing everything in software but gets eight times the routing performance for a 25% increase in clockspeed.
--
Developer: Tomato/MLPPP, Linux/MLPPP, etc »fixppp.org


RMerlin

join:2009-10-09
Montreal, QC
said by Guspaz:

So you're saying that 125% the clockspeed on a virtually identical CPU can produce 800% the routing throughput? That's a pretty extreme claim.

We're talking about benchmarks that show the N16 doing ~100ish and the N66U doing ~800ish here, and it's the same CPU architecture (literally the same core at a different clockspeed). I'm going to argue that it doesn't make any sense at all that the N66U is doing everything in software but gets eight times the routing performance for a 25% increase in clockspeed.

Was the RT-N16 benchmarked with the old 2.x FW, or with the newer 3.x firmware?


RMerlin

join:2009-10-09
Montreal, QC
BTW, without CTF, the RT-N66U WAN routing caps at around 200 Mbits based on some user tests. So I would expect the RT-N16 to also be able to break 100 Mbits if that module gets loaded, unless Broadcom hid some N66/AC66-specific magic in there. Gotta love closed source code...

morisato

join:2008-03-16
Oshawa, ON
Reviews:
·TekSavvy Cable
·TekSavvy DSL
·ELECTRONICBOX
reply to Optional
For me its partially a power concern partially i just would rather not use a pc for a router though i do have a Nice b/g/n card i could toss in my old case and it already has a pair of gigabit i still don;t particularly want to use it as a router.
--
Every time Someone leaves Sympatico an Angel gets its wings.

Doeboye

join:2006-11-07
Canada
reply to RMerlin
said by RMerlin:

Unlike their Ralink-based devices (like the RT-N56U), there isn't that much HW offloading on the RT-N66U.

...

I think the RT-N16 also has that same module, provided you use a recent firmware (both routers now share the same code base).

said by RMerlin:

BTW, without CTF, the RT-N66U WAN routing caps at around 200 Mbits based on some user tests. So I would expect the RT-N16 to also be able to break 100 Mbits if that module gets loaded, unless Broadcom hid some N66/AC66-specific magic in there. Gotta love closed source code...

Hi RMerlin,

What does this mean, performance-wise between the 56u and the 66u? Are you suggesting that the 56u is a better performer with QoS running?

Also, re: RT-N16: So if I understand you correctly, you are suggesting that if I retested with the newest Asus firmware, as opposed to the 2 year old TomatoUSB that I am currently running, I would see better results? More specifically, would I be able to exceed 150mbps?... If so, that would mean I could keep the existing router and save myself a few bucks when I upgrade to 150/10...


RMerlin

join:2009-10-09
Montreal, QC
said by Doeboye:

Hi RMerlin,

What does this mean, performance-wise between the 56u and the 66u? Are you suggesting that the 56u is a better performer with QoS running?

Also, re: RT-N16: So if I understand you correctly, you are suggesting that if I retested with the newest Asus firmware, as opposed to the 2 year old TomatoUSB that I am currently running, I would see better results? More specifically, would I be able to exceed 150mbps?... If so, that would mean I could keep the existing router and save myself a few bucks when I upgrade to 150/10...

Not sure, as I don't have an RT-N56U to test, and most benchmarks/reviews are also dating back to the older pre-Asuswrt firmware. But third party firmwares like Padavan do have a good list of Ralink-specific optimization options available, and these might not be incompatible with QoS unlike their Broadcom-based routers.

RT-N16: there's a good chance that running an up-to-date version of Asuswrt (or Asuswrt-Merlin) might give you better performance than Tomato. I'd say it's worth a try for sure, I can't tell for sure because once again most benchmarks are dated back to much older firmwares.

Doeboye

join:2006-11-07
Canada
said by RMerlin:

RT-N16: there's a good chance that running an up-to-date version of Asuswrt (or Asuswrt-Merlin) might give you better performance than Tomato. I'd say it's worth a try for sure, I can't tell for sure because once again most benchmarks are dated back to much older firmwares.

I'll definitely give this a try! I'm very curious to see if my WAN to LAN numbers go up!

Regarding the newest version of Asuswrt-merlin, you show it as beta for the RT-N16. Is it because the regular Asuswrt firmware (3.0.0.4.354) is also beta (issues with 3G/4g USB), or is it other issues that have been introduced?

Thanks!


Optional

join:2012-02-26
Mississauga, ON
Reviews:
·TekSavvy Cable
reply to Optional
I'm just a rank 4 nerd guys.. I dont know how to make my extra computer a router. It's not even reliable seeing as the psu is dying, it shuts off sometimes.

That asus n900 is a good bet yes? Everywhere I look people are drooling over it, it must be good.


RMerlin

join:2009-10-09
Montreal, QC
reply to Doeboye
said by Doeboye:

I'll definitely give this a try! I'm very curious to see if my WAN to LAN numbers go up!

Regarding the newest version of Asuswrt-merlin, you show it as beta for the RT-N16. Is it because the regular Asuswrt firmware (3.0.0.4.354) is also beta (issues with 3G/4g USB), or is it other issues that have been introduced?

Thanks!

This build includes a new wireless driver. Currently the RT-N66U driver has serious stability issues on the 5 GHz band - no idea if the RT-N16 might also be affected on its 2.4 GHz band. I'd recommend staying with 270.26b for now, the last stable version.

Doeboye

join:2006-11-07
Canada
said by RMerlin:

This build includes a new wireless driver. Currently the RT-N66U driver has serious stability issues on the 5 GHz band - no idea if the RT-N16 might also be affected on its 2.4 GHz band. I'd recommend staying with 270.26b for now, the last stable version.

Will do!

Doeboye

join:2006-11-07
Canada
reply to RMerlin
Update!

So, once the wife went to bed, I logged into my Asus RT-N16, flashed it with the latest stable version of AsusWRT-Merlin (270.26b), and re-hooked up my test WAN-LAN setup.

Same wiring/switch/ etc.

Here's the a link to my previous findings using TomatoUSB:
»Re: affordable router to handle 150/10?

Frankly, I could not believe the difference!

WAN to LAN used to be (TomatoUSB Firmware):
WAN to LAN: 11-13000 KB/s (86 - 101 mbps)
LAN to WAN: 12-15000 KB/s (94 -117 mbps)

WAN to LAN now (AsusWRT-Merlin Firmware):
WAN to LAN: 32-33000 KB/s (256 - 264 mbps)
LAN to WAN: 22-23000 KB/s (176 -184 mbps)

Unbelievable! Looks like this HW Acceleration feature really does something. I turned on QoS (Which seems to disable HW Acceleration), and my transfer rates dropped to around (actually a bit less) what I got using TomatoUSB...

So, conclusion: Guess I might be returning the new RT-N66U I just bought! . This router is now more than sufficient for the new 150/10 package we are going to be seeing (Assuming you don't care about QoS)!

Thanks RMerlin for the heads up on the firmware!

I honestly wouldn't have believed the difference if I didn't see it myself (and transferred files back and forth for a couple of hours).

technocar2

join:2009-05-29
Brampton, ON
kudos:2
How much was WAN to LAN and LAN to WAN with QoS on? I just can't fathom it being more than 100mbps; if you have time please do post it, I would really like to see the results with QoS.

simsin0

join:2008-01-06
reply to Optional
so is there any benefit for me to flash my rt-n16 from tomato-toastmanUSB to Merlin?

I only have 15/1 and plan to change to 50/10