D-Link Forums
The Graveyard - Products No Longer Supported => Routers / COVR => DIR-655 => Topic started by: marekwk on March 02, 2008, 07:36:23 PM
-
trying to replace my Linksys WRT54G. bought one of their "N" products, and was getting ~4.5 MB/sec as opposed to ~2.5 MB/sec on the "G" product. However, the router kept dropping connectivity and needed resettting few times a day. Since I was eyeing this D-link unit for awhile, Liksys went back as an RMA.
And today I picked up D-655. Set up was a breeze for the most part (although still need to spend some time on QoS). No dropped connections is a good thing. However, I am only getting ~2.8-3.5 MB/sec when transferring files to "N" laptop. Hardly an improvement :'(
Any ideas? Anything I missed in the setup?
802.1 mode = mixed
transmission rate = auto
channel width = auto (20/40)
security =WPA
there are 4 wired connections (gotta love 1MB) and 2 laptops, one "G" adapter and one "N". Before I bought this unit I asked whether intrduducing the "G" client would throttle down my "N" connection. Was told that's not the case.
-
I have a similar situation, previous wireless router/gateway was Linksys WRT45G,
then I purchased the DIR-655 and some DWA-552 PCI adapters to increase the
connection for a couple computers in the house.
I setup the DIR-655 as N only, and kept the WRT45G on the network for 'G' clients.
You need to make sure you disable all the 'gateway' features on the router that is
not handling your ISP connection.
I get 270 @ 100% signal on the DIR-655 (no drop-offs). The WRT54G operates at 54 (as always)
-
well, I would like to remove the WRT54G entirely. And the connections speeds are not really that relevant to me. I am more interested in "real life" improvement in transfer speed of files between clients. Guess I could get an D-link "N" adapter for the laptop that's currently using "G" and see what happens when I am using only "N" mode on the router.
cheers!
-
I've been seeing this. Look at the CPU usage. It may be spiking causing the LAN to LAN performance to degrade.
-
... And the connections speeds are not really that relevant to me. I am more interested in "real life" improvement in transfer speed of files between clients.
What method or program do you use to benchmark improvements on your "real life" transfer speeds?
-
transferring the same folder (different types of files) between the laptop and desktop using different routers. Consistent results w/ specific routers. Same PC / laptop hardware. No apps running... only variable are the routers.
-
transferring the same folder (different types of files) between the laptop and desktop using different routers...
What is the technique you use to make the transfer?
-
I suspect that my technique will be called unreliable. however, this is what I use in real life, hence this test - I open network neighborhood, locate the folder (always the same, on the same box), and then drag & drop to the same destination on the same target laptop. the "real" life scenario.
-
And today I picked up D-655. Set up was a breeze for the most part (although still need to spend some time on QoS). No dropped connections is a good thing. However, I am only getting ~2.8-3.5 MB/sec when transferring files to "N" laptop. Hardly an improvement :'(
i also just bought a d-655 router... have 2 wireless n setups (connected speed 270Mbps) and 1 hard wired (connected at 1 Gbps). when transferring from the wired computer to either wireless i see same as you around 3.5 MB/sec however when transferring from wireless to hardwired computer i get nearly double the speed at 6.5 MB/sec... am curious why the difference... ideally if it can be achieved in one direction it should be achievable in the other.
-
what wireless adapters do you have on your notebooks?
-
what wireless adapters do you have on your notebooks?
not running notebooks both are desktops located in other rooms... i am using draft 2.0 compliant adapters as follows:
AWLH6080 300N Wireless PCI Adapter
one wireless connects at 270 and the other at 300 both are running xp pro sp2
-
different results here... and not as good as yours. tried it in both direction:
1) from wired desktop (XP Pro SP2) to a laptop (Dell 1505 a/b/g/n adapter... still digging to find out who really makes it) running Vista Ultimate SP1 = ~3MB / sec.
2) From laptop back to the desktop (same folder) only getting ~2.5 MB / sec.
and yes, I am only connecting at 117 on "N" laptop and 54 on "G" laptop.
-
different results here... and not as good as yours. tried it in both direction:
1) from wired desktop (XP Pro SP2) to a laptop (Dell 1505 a/b/g/n adapter... still digging to find out who really makes it) running Vista Ultimate SP1 = ~3MB / sec.
2) From laptop back to the desktop (same folder) only getting ~2.5 MB / sec.
and yes, I am only connecting at 117 on "N" laptop and 54 on "G" laptop.
i see you're connecting at 117 and noticed from your 1st post you are running wpa not sure what cipher... however from a sticky post from lycan you need to be running the following encryption settings to obtain connection of 300:
"If you are using encryption the router must be configured to use AES cipher. In some firmware versions/models, the only way to get the AES cipher is to run WPA2. If you can not select the Cipher, selecting WPA2 only will force this cipher type."
confirming his position if you log into your router and go into manual wireless setup you will notice near the bottom of the page a discussion pertaining to wpa or wpa2 ending with the following recommendation:
"To achieve better wireless performance use WPA2 Only security mode (or in other words AES cipher)."
-
I used IOmeter http://sourceforge.net/projects/iometer to test my DIR-655 in N only mode
with a connection at 300, and I see ~4.5 MBytes/sec
-
Copying from a wired desktop to my laptop with an Encore wireless N adapter, 270mbps link speed... I see about 5-6MB/sec. Going from the wireless to the desktop, I see 6.5 and almost 7MB/sec. That's a bit strange, eh?
-
To do a proper performance test to see what you are really getting I would use a program called IPERF which is free :)
Also the other thing I would do is try changing the channel width to 20mhz instead of auto.
Also the Wireless G card should not throttle your connection down on your N card to something unbearable.
-
What we don't know as end users is the word N, we should also buy N devices so that you can get the max out of this N Gig router, am i right?
-
I am glad to see someone else analyzing this issue as well. I have done testing for more than two months and have yet to get better than 4.5 MB/s transfer rate and I have good programs to test them with. If interested look up TTCP on google. I have tried Linksys WRT350N and the DIR-655 with similar results (somewhat better on DLink). I have done EVERYTHING. The only other thing that I see different from yours that made a difference for me is sticking to wireless N only. I get slight improvement using that. I have been beating my head against the wall because 270mb/s divided by 10 calculates out to 27MB/s if all else is well.
Also, what happened to the promotion of 14x speed? If you got 2.5MB/s on G then 14X2.5=35MB/S, right. BTW, my test throughput on the router direct is about 26MB/s using Cat 6 cables. Needless to say, I am disappointed at the speed improvement on N but it is still about two times faster but very expensive to make that transition. Is it worth it, NO!
I hope that you find some other way to improve the speeds. I don't have any hair left to pull out. Good luck!!!!
-
I am glad to see someone else analyzing this issue as well. I have done testing for more than two months and have yet to get better than 4.5 MB/s transfer rate and I have good programs to test them with. If interested look up TTCP on google. I have tried Linksys WRT350N and the DIR-655 with similar results (somewhat better on DLink). I have done EVERYTHING. The only other thing that I see different from yours that made a difference for me is sticking to wireless N only. I get slight improvement using that. I have been beating my head against the wall because 270mb/s divided by 10 calculates out to 27MB/s if all else is well.
Also, what happened to the promotion of 14x speed? If you got 2.5MB/s on G then 14X2.5=35MB/S, right. BTW, my test throughput on the router direct is about 26MB/s using Cat 6 cables. Needless to say, I am disappointed at the speed improvement on N but it is still about two times faster but very expensive to make that transition. Is it worth it, NO!
I hope that you find some other way to improve the speeds. I don't have any hair left to pull out. Good luck!!!!
270mbit wireless does NOT mean 27MB/sec throughput. Period. It sucks that they market wireless that way. My wireless N router is about 2-3x faster than my old G one when I use wireless N and a bit faster in G mode as well. For less than I paid for my old G router, it is still a good deal. I get 4-7MB/sec thru wireless N.
-
270mbit wireless does NOT mean 27MB/sec throughput. Period. It [exploitive deleted] that they market wireless that way. My wireless N router is about 2-3x faster than my old G one when I use wireless N and a bit faster in G mode as well. For less than I paid for my old G router, it is still a good deal. I get 4-7MB/sec thru wireless N.
Using a DWA-552 and the DIR-655 it reports 270 or 300 mbps connections. But pretty much like the B->G upgrades reality is less good. Some argue that reality-based benchmarks aren't valid -- and of course that's about as far as the truth as one can get. One gets what one gets in actual use and little else really matters. Real use is the absolute best benchmark. What we need the speed for is medium to large file transfer to/from a fileserver (on the LAN) to the wireless machine and I used the windows task manager networking graph to measure throughput (the 552's manager is pretty spartan, most other ones I've had in the past usually have nice little throughput graphs and the like).
We get 2~3x throughput improvement (over previous 'G' system) under the conditions above. It's about what I expected and am satisfied even if not thrilled.
-
In an wireless N environment, using a salvaged N650 radio from a dead DIR-635 I was able to upgrade my Tablet PC to N. With that N card in there I was able to use a Tablet with a 900Mhz cpu and a base Intel graphics adapter to stream HD movies. When benchmarked with Chariot, I saw speeds of up to 32MBps. (300 mbps connection rate).
Now, some would say that I cheated because I used a radio from a router and not an adapter, but that just leads me to believe that the reason for poor wireless performance in N mode is a combination of Drivers, the systems running the cards and the manufacturer of the radios themselves. The technology is sound, it's just an infant. These are the growing pains of Wireless N.
-
In an wireless N environment, using a salvaged N650 radio from a dead DIR-635 I was able to upgrade my Tablet PC to N. With that N card in there I was able to use a Tablet with a 900Mhz cpu and a base Intel graphics adapter to stream HD movies. When benchmarked with Chariot, I saw speeds of up to 32MBps. (300 mbps connection rate).
Now, some would say that I cheated because I used a radio from a router and not an adapter, but that just leads me to believe that the reason for poor wireless performance in N mode is a combination of Drivers, the systems running the cards and the manufacturer of the radios themselves. The technology is sound, it's just an infant. These are the growing pains of Wireless N.
If you are getting 32MB/sec with your hacked setup, why don't you just sell that to us? That's more than six times what I get.
-
In an wireless N environment, using a salvaged N650 radio from a dead DIR-635 I was able to upgrade my Tablet PC to N. With that N card in there I was able to use a Tablet with a 900Mhz cpu and a base Intel graphics adapter to stream HD movies. When benchmarked with Chariot, I saw speeds of up to 32MBps. (300 mbps connection rate).
Now, some would say that I cheated because I used a radio from a router and not an adapter, but that just leads me to believe that the reason for poor wireless performance in N mode is a combination of Drivers, the systems running the cards and the manufacturer of the radios themselves. The technology is sound, it's just an infant. These are the growing pains of Wireless N.
Lycan,
Uhhhhh, I don't buy it. 32MBps x 8 bits = 280 Mbps. At 300 mbps connection rate that's 93% efficiency. That efficiency would on par for a full duplex Wired connection not a half duplex wireless connection. It's just not going to happen in the real world dude. Try giving us your application data rate not your signaling rate. Then take your setup out of the lab and into the real world where you have to contend with RF interference from all the numerous sources.
Bah humbug,
bspvette
-
Lycan
Just rechecked my throughput between a couple laptops with DWA-652 cards and my DIR-655 after updating to firmware 1.20. I am still consistantly getting 70-80 Mbps Data throughput with a 300Mbps connection rate. I'll be generous and say 80 Mbps. Devide that by 8 and that is roughly 10MBps. 80 Mbps / 300Mbps = roughly 27% data throughput efficiency.
Now for fun, we plug the laptop with a 100 Mbps full duplex ethernet connection into our DGS-2208 switch and run the exact same test between the same systems (the other sys has Gb). Results: 93Mbps continuously. Thats 11.6 MBps and a data throughput efficiency of 93%.
For even more fun, we run the same test between two systems with GB NICs... NOW we are getting in the 250-350Mbps range using either the DGS-2208 Gb switch or the DIR-655. (might retry this with some real Gb cables)
Hence, therefore, thus, Wireless-N can barely compete with 100BaseT and doesn't have a prayer against 1000BaseT.
It also appears you were using a WIRED Gb connection for your tests.
Cheers!
BSPvette
-
Hey, 10MB/sec through 802.11n is great. Competing with fast ethernet is the real goal for me.
-
Hey, 10MB/sec through 802.11n is great. Competing with fast ethernet is the real goal for me.
EricP,
If you have only two devices talking and no interference, you can get close to that goal; otherwise, you would be wise to stick to switched, full duplex, and wired. I think alot of people think going from wireless G to N is going to give them the same 10x or 100x increase they saw going from 10Baset to 100Baset or Gb. The best that could be achieved for N would be a 5.6X increase over G. (300/54=5.555555)
Cheers
BV
-
I am using a $15 encore 802.11n adapter in an 802.11NG environment with multiple devices and multiple wireless networks in the area. I walked the laptop with the 802.11n adapter up the stairs and sat right next to the router and maintained a 270-300mbps connection the entire time but throughput never goes above 7MB/sec :(
-
what are you using to judge the throughput speeds?
-
what are you using to judge the throughput speeds?
Lycan,
I am using ixia qcheck freeware edition: http://www.ixiacom.com/products/performance_applications/pa_display.php?skey=qcheck
Throughput test with 1000KB Data size
Do you have a link to the product you are using so we can use the same tools?
Regards,
BV
-
Lycan,
I am using ixia qcheck freeware edition: http://www.ixiacom.com/products/performance_applications/pa_display.php?skey=qcheck
Did you find it difficult to get the qcheck freeware? I tried using their form to get a copy (twice)
and I get zero response from them. What's funny is that my day-job is at a company
that probably could have used their mainline software (if we don't already somewhere) so their
non-sending of the download link has been particularly un-impressive. Does qcheck work well?
If not I'll be in less of a huff in their non-response. :-)
-
Oregonian2,
I downloaded it a few months ago with no problems when I was truly getting bad throughput via wireless with my DIR-655 and DWA-652. (sometimes < 1Mb) related thread: http://forums.dlink.com/index.php?topic=1268.0
I can't speek for their web site today but it has a been a decent tool for my home use.
Regards,
BV
-
I was able to register an email account today. It displays the link to the file as
http://www.ixiacom.com/downloads/products/qcheck/qcinst3.0.exe but has tracking info in the actual link....
Regards,
BV
-
I was able to register an email account today. It displays the link to the file as
http://www.ixiacom.com/downloads/products/qcheck/qcinst3.0.exe but has tracking info in the actual link....
Regards,
BV
Thanks for the link! They sent me email rejecting my request saying that they don't accept
email addresses like those from yahoo or gmail and that they don't see me as a registered
owner of their products (such as Chariot). I had given them my email address at
work (a 30+ year old good sized company in the telecom business, but not one
that competes with them afaik -- a potential customer type) as well as my own
personal domain-name email address.
Anyway I'm not sure how you got the software, but they have something
against those in Oregon or maybe just me. But needless to say I'm not too
thrilled with them and the likelihood of my recommending their products
to our testing department (where most likely used -- I'm in engineering) is pretty nil. But at least
I got a nice form rejection letter rather than just being ignored completely.
But anyway, apparently one has to be "proper" enough to qualify for their
free software -- presumably those like me would do something dastardly
with it. Or it's not really free, kinda depends how one looks at it.
-
That's really odd because I used a yahoo email address to register and they sent me a valid download link.
Oh well. Their bad!
BV
-
The thing that's really funny is that after rejecting my request for their free Qcheck software, they've now started sending me promotional email (with the obvious intent of trying to get me to buy things from them). Talk about optimists ! Needless to say I've requested to be taken off of their mailing list. I hope they grant this request (I have hope, it appears to be an automated system).
:-)