Useless War Boards

[patos]

Lonnie,

We got the famious WAR boards and apparently they are useless on our network . We are unable to run the RIP on them ( we have few hundreds routes ) and even the static route more than 3 is forbidden. Plus turbo mode not working same speed as 54 rate.

Plus the Wrap boards also a failure , after years of development still they reboot and even GOD knows why . The PC based atheros is a TOTAL FAILURE , with Atheros , crashing all the time .

Please i would like before i can take any steps , when all this will be fixed .

Thanks in advance.

[lonnie]

It has to be remembered that when doing a throughput test yopu must do it through the unit, and not on the unit. There simply is not enough CPU power, even on the WAR boards, to do a bandwidth test and have the test software running on the machines being tested.

If you do a test from a PC to a PC through the WAR boards you will see 8 MBytes/sec or more depending on your local situation.

We did discover a memory limit to RIP routing that only accepts 1,016 routes. You have 170 Access Points so I suspect you are overflowing the route table. We were about ready to release the trial image, but with your recent post it is obvious we must do some more testing. I would not want to disappoint you again.

As for the WRAP and PC reboots, that depends on the territory. It is a combination of power and traffic. Simply put, we, as do many others, have WRAP and PC runtimes measured in months, so in general there is no issue with hardware or software.

The best installs are obtained with cat5 punchown equipment. I sent you some samples of the new toolless units we are using. If you crimp a RJ45 onto a cat5 solid cable you do not have a proper termination.

We also have the best luck with battery powered systems. There is far less noise and runtimes are the longest. We do have some PC units as servers but they are on a huge UPS system.

[patos]

Lonnie,

Thanks a lot for the credible information . I will look into it . As for the RIP i am trying to minimize all routes so i think i will be able to run them finally on the WAR boards .

But concerning the WRAP and PC crashing or rebooting . I know all what you saying is correct but , I have this issue only on the AP mode , the station mode or client mode connected to the AP that is crashing or rebooting are for monthssssssssssss on and on without any reboot. I repeat that again only AP side crashing . SO it is not any cable or power issues . Yes i see better runtimes with batteries , we always use them . But the issue whatever the situation on the client side they are alwayssss uptime . SO only AP side is the issue .

Thanks in advance .

[patos]

Lonnie,


For more accurate info we have excatly 600 routes in the RIP . But we are finalliy applying route summarization , So we will have much lessssss routes in RIP .

Thanks in advance .

[patos]

Lonnie,

For more info also , i did speed test from :

PC P4 >> WAR on turbo -60 signal >> WAR distance 3 KM >> P4

speed before turbo was 4600k/s
speed with turbo 5000 k/s .
So as i see no difference in speed . Or maybe limited by CPU .

I am sure that if i will have 2PC based atheros i will see 10.000k/s i have 2 similar sites , i see daily this speed on P4 PCs .

In My opinion this WAR boards are only twice the speed of the wrap boards.

Thanks in advance .

[phendry]

PC P4 >> WAR on turbo -60 signal >> WAR distance 3 KM >> P4

speed before turbo was 4600k/s
speed with turbo 5000 k/s .
So as i see no difference in speed . Or maybe limited by CPU .
Just out of interest, are these WAR boards the 2 port (266 CPU) or 4 port (533 CPU)?

[mp3turbo]

> I am sure that if i will have 2PC based atheros i will see 10.000k/s
> i have 2 similar sites , i see daily this speed on P4 PCs

patrick, will you elaborate more on this? How do you get 10MB/s thru Atheros? Compressed data? Is it filetransfer test with zip files or built-in data transfer test?

[tog]

Sounds like the 2-slot WAR board with the 266MHz CPU.

5000K/s through it is where mine tops out, too. Not too shabby, especially considering nothing we're doing out there in the real world is as compressible as the all-zeros file I use to get up to 5000K/s. 5000K/s possible system throughput for running two radios is quite acceptable.

I get over 10,000K/s on PCs running StarOS using an Atheros CM9 in turbo mode transferring compressible data. :)

http://forums.star-os.com/showthread.php?t=3146

I still love the VIA EPIA more than anything else ever.

[patos]

> I am sure that if i will have 2PC based atheros i will see 10.000k/s
> i have 2 similar sites , i see daily this speed on P4 PCs

patrick, will you elaborate more on this? How do you get 10MB/s thru Atheros? Compressed data? Is it filetransfer test with zip files or built-in data transfer test?


It is the built in transfer of Staros

[patos]

Just out of interest, are these WAR boards the 2 port (266 CPU) or 4 port (533 CPU)?

It is the 4 port 533 WAR boards. I did test also from PC to PC bypassing them without turbo i see 5000 , with turbo a little bit more like 5400 but this is routed and not bridged.

[patos]

Well guys,

I already spoke with lonnie about this.
We are an ISP with 25000 users online . So we have the most weird experience in the world with DSL , wireless and anything .

Well when u see 5000k/s on ur wireless this is from PC to PC txt file that can be compressed to the maximum and the most important the MTU set to 1500 by default.

Well the internet world is different where max average MTU is arround 400 .
So if u will put the internet traffic on ur 5000k/s link when tested u will get 4 times less arround almost 1500 k/s .

We invent our own system working on linux it is imilar to VDS in Staros .

And when we tested on a 54MG/s wireless link on Staros test i was getting 5000 k/s with no one on it . When used with our system i got 4000k/s and there was 30 mg/s traffic on the link . Without our tunnel system the 54mg/s tested as 5000k/s was handling the 30mg/s internet traffic with i repeat 400 MTU average and we were getting packet losses . When we run our system we got 30 mg running to the other side no packetloss at all . And able to load the wireless by 4000k/s over the 30mg/s already loaded.

We did hard test and we reached 91 mg/s speed on the 54mg wireless link and 119mg/s on the 108 wireless link .

Already i ask Lonnie to code this is Staros, it will be the gadjet of the year . No one have this on the planet . But bad news it is not for free.

Thankd in advance .

[phendry]

I'm guessing this is something similar to Mikrotiks M3P where it bundles a load of small packets into 1500 byte packets. If you then tunnel this across your network you have end to end large packets. This is exactly what we do but I haven't tested it to see what speeds we can get across a StarVX link yet.

[patos]

I'm guessing this is something similar to Mikrotiks M3P where it bundles a load of small packets into 1500 byte packets. If you then tunnel this across your network you have end to end large packets. This is exactly what we do but I haven't tested it to see what speeds we can get across a StarVX link yet.


Yes it is similar 99% for Microtik . But but but but what if ur link have 40% packet loss ?? . With our system even if u have 40% packet loss , u will not see one packet loss the other side of tunnel , the only thing u will see a little change in jitter few ms more . And also on top of that our own compression .

P.S: Hearing is not like seeing !!!

[phendry]

Isn't bundling small packets into big packets what Atheros's "Fast Frame" is supposed to do? Maybe the code isn't working correctly as we do see a big impact when testing with small frame sizes across a CM9 to CM9 link.

[RomaWireless]

Looks like the tecnique you mentioned has found an usage in your satellite service ... BTW interesting south-european footprint ...

Hey, I can try to convince the boss, would you share transpoders costs so we peer over panamsat 4? :)
At least, can I receive it with the a dreambox ?

nothing is more wireless than satellite afterall :)

[mp3turbo]

hi patos,

> We did hard test and we reached 91 mg/s speed on the 54mg wireless link and 119mg/s on the 108 wireless link


I guess you had very compressible data here in your test, did you? It is not possible to transfer 91Mbit/s of ZIPped file thru 54Mbit/s WIRELESS datarate (which is even HALF duplex) no matter what tunnels/encryption you use. And the same applies to 119Mbit/s thru 108Mbit/s wireless... if you were testing with compressible data, than this is not important - you can transfer 300Mbit/s thru 36Mbit wireless datarate... when you have "enough compressible text file".

Can you provide more details about your solution?

thnx!

[patos]

hi patos,

> We did hard test and we reached 91 mg/s speed on the 54mg wireless link and 119mg/s on the 108 wireless link


I guess you had very compressible data here in your test, did you? It is not possible to transfer 91Mbit/s of ZIPped file thru 54Mbit/s WIRELESS datarate (which is even HALF duplex) no matter what tunnels/encryption you use. And the same applies to 119Mbit/s thru 108Mbit/s wireless... if you were testing with compressible data, than this is not important - you can transfer 300Mbit/s thru 36Mbit wireless datarate... when you have "enough compressible text file".

Can you provide more details about your solution?

thnx!

Dear ,

We are an ISP with 23,000 wireless users . And when i am talkign about transfer rate , I am talkign about our backbone transfered from Loacation A to B . I am not a PC sending a ZIP file to another PC at the second room . So all my numbers are related to an extensive Internet traffic , . If you are not an ISP this is not good for YOU .

And As for Mr RomaWireless Lover , we are using same technic on Panamsat 4 as u mentioned and my DVB carrier is 45 mg and i am getting 60 MG after the DVB receiver .

Once again this is Internet Traffic not a ZIP file .
45 mg/s cost 45.000$ / month over SAT . SO i am doing 15 mg for free = 15.000$ .

[phendry]

Hey Patos,

Did you ever persuade Lonnie to include your code? I'm very interested in trying it. Does it add delay like Mikrotiks M3P does?

[patos]

Hey Patos,

Did you ever persuade Lonnie to include your code? I'm very interested in trying it. Does it add delay like Mikrotiks M3P does?

Hi ,

Yes we spoke with Lonnie several time it is 100% positive , but as for the time being they are too busy releasing the V3 , so not the good time . As for our code it is a Unique code almost no one has ot , it is much different than what Microtik have it . There is one single product similar to us : www.mentat.com aquired by Packeteer now .

[phendry]

I have seen the packeteer product before but it's a bit pricey and it's difficult to deploy it in all locations. Have you considered releasing your proprietary software?