I've got a couple of links that I would like to start using Mikrotik on at between 5-15 miles each.

I'm currently getting about 25Mbps from a Proxim link using about 25Mhz of spectrum.

Is this possible to do with a RB500 series, two 802.11a cards and NStream or NStream 2?

Is dual polarized antenna better/worse than two seperate antennas?

I would like to stay in the 5.2GHz spectrum and possibly migrate to the 5.4GHz spectrum if it ever becomes legal to do so in the USA.

To get 60Mbps using an 802.11a link in Turbo mode, you'll need PC hardware. Someone on here posted some speed tests with something like a 2GHz Celeron showing 60Mbps. (40MHz of spectrum)

Using NStreme2 you should be able to pull this off with RB500s. However, you'll need at least 40 MHz of spectrum (2x20MHz) or 80MHz if you plan on using Turbo mode. Mikoritk recommends at least 200MHz of channel separation, so you'll need 5.2GHz and 5.8GHz or 5.4GHz when it becomes available.

In all cases you'll be using more spectrum than your Proxim link.

So with NStream2 I should be able to get something near 45Mbps actual throughput?

200MHz separation is hard and I don't want to dig into our 5.8GHz spectrum.

What antennas are good for the dual polarized scenario in the 5.2GHz range?

We are using Radiowaves sectors and Andrew dishes and drums.

I've got a couple of links that I would like to start using Mikrotik on at between 5-15 miles each.

I'm currently getting about 25Mbps from a Proxim link using about 25Mhz of spectrum.

Is this possible to do with a RB500 series, two 802.11a cards and NStream or NStream 2?

Is dual polarized antenna better/worse than two seperate antennas?

I would like to stay in the 5.2GHz spectrum and possibly migrate to the 5.4GHz spectrum if it ever becomes legal to do so in the USA.

From our testing, we've found that the most "optimum" configuration on a SBC (RB5xx, Wrap, etc) is point-to-point WDS bridging (not enough horsepower to realize Nstream improvements) -- in this mode, on a stock 20 Mhz channel -- you should expect to see a little bit over 25 Mb of aggregate TCP throughput


Stephen from Cable Free Solutions claims that using 1+ GHz PC-based processors & N-stream gives 30-35 Mb of aggregate TCP throughput on that same 20 Mhz channel (we are currently looking into trying to duplicate his results)

-Charles

Claim or reality?

Using Turbo mode (40MHz channel)
Screenshot of NMS from full-speed lab testing, 83Mbps UDP traffic with ~20% CPU load
http://www.cablefreesolutions.com/radio ... %20UDP.png
Screenshot of NMS from full-speed lab testing, 74Mbps TCP/IP traffic with ~20% CPU load
http://www.cablefreesolutions.com/radio ... %20TCP.png

Those results some time back, using an older 2.8 version. MT have completely new drivers now in 2.9.
Obviously "real traffic" is somewhat worse, that was test traffic using routers and bandwidth-test.
Actually, on the other list, which Charles is quoting, I think I said 30-35Mbps UDP for 20MHz channel.
We do have complete set of current test results for 5, 10, 20, 40MHz channels in UDP and TCP but the're in some huge word file somewhere. Must make a web version soon.
I should mention as an MT OEM we have complete, outdoor-grade, passively-cooled, fully supported, shipping product. It will take 5 radio cards and we've demonstrated 200Mbps bridged LAN-wireless split between 5 interfaces in P2MP tests, we have both 10/100 and Gig-E ports on our boxes. But this is MT's support forum, not an advertising board so best contact us off list.
We also have a test LAN in our offices that (serious) prospective customers can remotely access, and try configurations for themselves, to show there's no myth here ...

Regards

CableFree Solutions

Claim or reality?

Using Turbo mode (40MHz channel)
Screenshot of NMS from full-speed lab testing, 83Mbps UDP traffic with ~20% CPU load
http://www.cablefreesolutions.com/radio ... %20UDP.png
Screenshot of NMS from full-speed lab testing, 74Mbps TCP/IP traffic with ~20% CPU load
http://www.cablefreesolutions.com/radio ... %20TCP.png

Those results some time back, using an older 2.8 version. MT have completely new drivers now in 2.9.
Obviously "real traffic" is somewhat worse, that was test traffic using routers and bandwidth-test.
Actually, on the other list, which Charles is quoting, I think I said 30-35Mbps UDP for 20MHz channel.

Out of curiosity -- if you're getting 83 Mb UDP / 74 Mb TCP on a 40 MHz channel -- wouldn't it be logical to assume that you get 1/2 the throughput on a 20 MHz channel (e.g., ~41 Mb UDP / ~37 Mb TCP) -- if you're getting different results -- would you mind explaining why you think that might be the case?

-Charles

received-packets-per-second: 4212
received-bits-per-second: 32.9Mbps
sent-packets-per-second: 3972
sent-bits-per-second: 33.1Mbps

Real traffic on max. T-bird 1,4 and Celeron 1,6 @ 30 % CPU. WDS bridged, nstream; -60 dBm Tx Strength; 10 kms.

Update:
same system dual feeds:

received-packets-per-second: 9238
received-bits-per-second: 47.0Mbps
sent-packets-per-second: 8800
sent-bits-per-second: 44.8Mbps

The dual isn't the best:
wifi1:
signal-strength: -72dBm
tx-signal-strength: -72dBm
noise-floor: -99dBm
signal-to-noise: 27dB

wifi2:
signal-strength: -54dBm
tx-signal-strength: -61dBm
noise-floor: -108dBm
signal-to-noise: 54dB

The conf is:
2 EOIP bridges with nstream/best-fit in bonding-mode.

I'm looking for bonding improvement in future ! As I can reach around 70 mbit with one link and around 40 mbits with the other .

I've got a couple of links that I would like to start using Mikrotik on at between 5-15 miles each.

I'm currently getting about 25Mbps from a Proxim link using about 25Mhz of spectrum.

Is this possible to do with a RB500 series, two 802.11a cards and NStream or NStream 2?

Is dual polarized antenna better/worse than two seperate antennas?

I would like to stay in the 5.2GHz spectrum and possibly migrate to the 5.4GHz spectrum if it ever becomes legal to do so in the USA.

Look at this thing i posted, hardware is posted too. On smaller links we use the 5,2-5,3 band with cm9

http://forum.mikrotik.com/viewtopic.php?t=10924