150+ Mbps with 10/100 Ethernet Port

[etijburg]

Can someone explain to me how it is possible to provide 150+ Mbps to Subscribers with only a 10/100 Ethernet Port?

[WHT]

150 Mbps is the half-simplex radio data speed, so effectively its a 75 Mbps full-duplex ethernet speed.

[jesterz]

Here's a link to that topic covered earlier just reinforcing WHT's comments here...

http://www.ubnt.com/forum/viewtopic.php?t=13274

[Ron]

150 Mbps is the half-simplex radio data speed

You mean half-duplex.

[UBNT-Mike.Ford]

150 Mbps is the half-simplex radio data speed

You mean half-duplex.

This is correct.

Thanks,

Mike

[WHT]

No...half-simplex if you use that definition common to radio communications.

Duplex - Two different frequencies.

Simplex - One frequency

Full-Duplex - Simultaneous transmit and receiver, possible only with a full duplex arraignment.

Half-Simplex - Receive and then transmit, used on simplex radio setups like a "store and forward" repeater.

[WHT]

Mike...can you clarify what you said was correct?

150 Mbps is the half-simplex radio data speed.

or

You mean half-duplex.

[Ron]

No...half-simplex if you use that definition common to radio communications.

Duplex - Two different frequencies.

Simplex - One frequency

Half-Simplex - Receive and then transmit, used on simplex radio setups like a "store and forward" repeater.

Your definitions are wrong.
See http://www.ligaturesoft.com/data_communications/communication-types.html

"A simplex system is a communication system in which the message can be send in one direction only."

"In a half duplex system, each end may transmit, but only one at a time. This requires both transmitting and receiving circuitary at each end, but the actual link between the two ends may be shared.

Eg : A citizen’s band radio where a frequency channel is shared and each party has to say “over” to switch the direction of the communication. "

So, again, 802.11 is half duplex.

[WHT]

My definition is correct when using the recognized definition used by radio professionals.

A CB radio is half-simplex as a user can only talk half the time. Its simplex as it uses the same frequency.

[UBNT-Mike.Ford]

Mike...can you clarify what you said was correct?

150 Mbps is the half-simplex radio data speed.

or

You mean half-duplex.

When I refer to 150Mbps, this is 150mbps TCP/IP Half Duplex speed, not radio operation.

Thanks,

[WHT]

So you are using the phrase to describe it as talking half the time in one or the other direction and which would the transport over a single frequency.

OK, that's one alternative way of describing it.

[jks19714]

Marketing-speak Data Rate = Goes-Inta's + Goes-Outas + Round_to_Whatever_Sounds_Good

It's Math_for_Marketers101, required in every b-schools' MBA program. :roll:

If real engineers did math that way, water wouldn't pass under bridges and the Space Shuttle would be permanently affixed to the launch pad.

john

[UBNT-Mike.Ford]

Marketing-speak Data Rate = Goes-Inta's + Goes-Outas + Round_to_Whatever_Sounds_Good

It's Math_for_Marketers101, required in every b-schools' MBA program. :roll:

If real engineers did math that way, water wouldn't pass under bridges and the Space Shuttle would be permanently affixed to the launch pad.

john

The numbers arent made up.

Regarldes off half duplex, half simplex, full duplex ect terminology, you can see up to 150Mbps of TCP/IP (with 300Mbps over the air rates) going over the link.

Thanks,

Mike

[osnet]

Man why the people don't do a search in the forums.... Always come the same question again and again....

[Blizz]

half simplex or half duplex, in term of delivering information data rate from point A to B it doesnt even matter, the guys just want to explain that it would transmit 150Mbps over that interface, plain simple... :)..

[WHT]

With artistic license in my explaination...

During the first second, 150 Mb is sent to the far end. During the second second, 150 Mb is echoed back to teh near end for data confirmation.

So over a period of two seconds, 150 Mb of data is sent and confirmeed back.

150 Mb per two seconds is 75 Mb per second.

[Ron]

During the first second, 150 Mb is sent to the far end. During the second second, 150 Mb is echoed back to teh near end for data confirmation.

So over a period of two seconds, 150 Mb of data is sent and confirmeed back.

150 Mb per two seconds is 75 Mb per second.

"Data confirmation" is obviously far fewer bits (like 90% less) than the data itself.

What are you trying to say anyway?
Mike already said it's 150 Mbps per second, not 75 Mbps per second.
75 Mbps is what I get on 802.11a.

[george]

There is one scenario where you can run out of steam on the Ethernet port.

If you have a short link, so you're getting the full radio speed, along with a highly asymetric data flow (which we have) you can overload the Ethernet port in the downstream direction while still having radio capacity left over.

We run about 8:1 on some APs, which would mean about 133Mbits/second on the downstream with only 16Mb/s on the upstream.

Obviously this isn't going to work too well on a 100Mbit Ethernet port. Does it matter? No, because we are extremely unlikely to be able to saturate the radio side effectively.

I think this design is a very nice balance.

George

[mhammett]

Marketing-speak Data Rate = Goes-Inta's + Goes-Outas + Round_to_Whatever_Sounds_Good

It's Math_for_Marketers101, required in every b-schools' MBA program. :roll:

If real engineers did math that way, water wouldn't pass under bridges and the Space Shuttle would be permanently affixed to the launch pad.

john

Only in unlicensed. In the licensed world, they only rate it as one of the "Goes-XXXs". Rounding still applies.

[SMART]

the numbers arent made up.

Regarldes off half duplex, half simplex, full duplex ect terminology, you can see up to 150mbps of tcp/ip (with 300mbps over the air rates) going over the link.

Thanks,

mike

mike: The m products is fulduplex or half duplex please clarifi

thanks.

[Dave-D]

All Ubiquiti radios are half-duplex,
as Mike as said many times. Dave

[mhammett]

mike: The m products is fulduplex or half duplex please clarifi

thanks.

The search feature must be broken.

The radios are half duplex as are all UBNT radios. The Ethernet is full duplex.

[lncommunications]

http://learn-networking.com/network-design/carrier-sense-multiple-access-collision-detect-csmacd-explained

[mhammett]

I saw it go over 110 megs at the AirMax Conference Chicago. They say that it'll do 150 on a PtP and 100 on a PtMP.

[NZFoxnet]

I see the ethernet saturate at 97.4Mbps UDP on a lot of my NSM5 links.

[InoX]

110mbps TCP on NS5M's

[sakita]

This one'll probably never die... (e.g. see http://www.ubnt.com/forum/showthread.php?t=15016).

There's the over-the-air-rate that symbols are processed. Each symbol represents a bit or number of bits. There is some overhead from MAC, CRC, and other whatnots that result in actual usable data being transferred. Throw a little random interference into the mix and what you end up with is a range of rates. The range can include a theoretical (calculated) maximum as well as a 'to be reasonably expected' number (e.g. with nominal interference).

So, in the end it is a bit confusing (no pun intended). :icon_eek:

Whatever you want to call it - the radio devices at each end talk one at a time (or at least have to for it to work). If there is, for example, a capacity of 300Mbps over the air it can get sliced into upstream or downstream in various proportions.

The wireless is connected to the rest of the world through a copper port that can run both directions at the same time... 100Mbps in each direction.

Is it feasible that with most of the data going over the air in one direction that one direction of the copper port could be saturated? I'd say yes. In the real world is this likely to be a problem? Probably not. A 100Mbps copper port looks to me like a reasonable design.

As for the data rate it does get extra confusing because some manufacturers aren't clear whether they are talking symbol rate, bit rate, or whatever. They way Ubiquiti has stated it is clear. They've stated a TCP (payload) rate. So, you can forget worrying about what is going on inside the radio with symbol rates, etc. So, 150+Mbps of payload can be crammed through.

Don't believe it? Try it for yourself or search for confirmation. Nothing short of phenomenal on a cost for rate basis really.

[opampca]

I saw it go over 110 megs at the AirMax Conference Chicago. They say that it'll do 150 on a PtP and 100 on a PtMP.
I achieve 120 Mbps steady at 10 meters distance in my backyard using 2 instances of Jperf back and forth with 10 connections each...pretty impressive, and I imagine I could get better if I could lower more the radio power (actually around -24dbm with radio at 8dbm).
Richard

[UBNT-Mike.Ford]

OK... Since this boils down to a little misunderstanding, but also a bunch of "Overglorifying" of the real values...

What is the max data amount that can be pushed through these radios in one single direction TCP at a time VIA Ethernet?? (No over the air values)

From what I see on a nicely tuned link, I can crack off around 73Mbps in a single direction TCP Ethernet. This theoretically is the max I can see in such a scenario correct? When I slam both directions evenly this gives me 45Mbps on each direction... So my numbers don't exactly line up, that 45 I think is a little off.

I don't like this 150Mbps claim because it's basically adding up both together. It's like saying an 802.11b link is doing 22Mbps. It's only 11 in each direction (Over the air yes I know I'm not comparing apples), but it's not right to say it's a 22Mbps radio.

I'm also tired and not thinking right, so someone please correct my math here if I'm off.

Given if my logic is right, and for once I hope it's not... These radios are 75Mbps radios and not the claimed 150. And for some reason I remember back in the infantile stages of these people asking if they were full duplex and some people were saying they were. When they are obviously not.


Hello,

You can get around 85Mbps TCP/IP in a single direction and this is limited by the Ethernet port. However, passing traffic in both directions you can get 85Mbps each way, so 170mbps TCP/IP AGGREGATE throughput.

So no, our claims are not false, you can see people getting well over 100mbps TCP/IP traffic aggregate. Once again, all 802.11 based radio's are HALF duplex, and speeds are measured HALF duplex. 170Mbps TCP/IP HALF Duplex is 85 Mbps TCP/IP in EACH direction at the same time, hence the aggregate term. Yes you are actually able to push this amount of data. These Radio's are NOT FULL duplex (wich would be 170Mbps TCP/IP in EACH direction at the same time.). 170mbps TCP/IP FULL duplex = 340Mbps TCP/IP HALF duplex.

We are not misleading anyone in any way as we have stated from the beginning that our radio's are half duplex radio's and that ALL results for maximum speeds are shown in half duplex, as evidenced by our chariot graphs showing bi-directional traffic.

Thanks,

Mike

Mike

[sakita]

Since (based on the above post) UBNT has IxChariot as one of their testing tools I'm wondering if anyone else here (in addition to jperf and other tools) uses the IxChariot freebie Qcheck? Being a stripped down product it does have limitations but I've found it to be handy. :icon_smile:

[opampca]

The thing I don't like with QCheck is the limitation of only one connection.
We usually get more throughput with multiple connections over the air.
JPerf is my prefered tool.
Richard