I have a question about ixchariot, I want to make test on TP-link tl-wr941ND using ix

[loopdemack]

I have a question about ixchariot, I want to make a test on the TP-link tl-wr941ND using the ixchariot.

100mbit wan-to-lan and vice versa environment.

Both network cards connected to wan and lan were 1gbit, but they are both selected in a 100mbit full duplex.

When I measured the total simultaneous connections using a higher number of pairs, I get some strange error about the timing, tell me which timing did you selected when you tested routers on the 100mbit network wan/lan.

And which script did you selected, and which quality of service did you used in the ixchariot, and did you changed(altered) something in the script.
And did you altered anything in the ixchariot basic settings, I'm using IXchariot 6.7.

The error was:
Error: 'CHR0336: A timing record was received with a measured time between 1 and 20 milliseconds.'

Tell me how do you know that you reached the maximal amount of total simultaneous connections, how could I recognize when I reached the pairs limits?

I did read your explanation, http://www.smallnetbuilder.com/content/view/26563/53/
but could you explain a little bit more. Thank you.

Could you explain me how could I tweak the timing that I could make a good environment for the 100mbit test, I want to overcome that error that I got.
I found in a http://www.ixiacom.com/support/ixchariot/knowledge_base.php ixchariot knowledge base about this error but they didn't told how to set timings according to network speed.

The Router I'm testing is TP-Link tl-wr941ND Atheros chipset MIMO
CPU: Atheros AR9132-AC1E 400MHz
Switch: Marvell 88E6060-RCJ1
Memory: HY5DU561622FTP-5
Flash: SPANSION FL032A1F
Radio: AR910

I want to find max number of pairs on this router.

 

[thiggins]

It is common to see that error message. It is usually due to too small a test file size. I don't worry about it.

I use throughput.scr, TCP with all defaults except changing test file size. For throughput tests with fast routers, I usually set the test file size to 3000000 Bytes. For max simultaneous connections, I drop that back to the default 100000 Byte file for all pairs.

 

[loopdemack]

It is common to see that error message. It is usually due to too small a test file size. I don't worry about it.

I use throughput.scr, TCP with all defaults except changing test file size. For throughput tests with fast routers, I usually set the test file size to 3000000 Bytes. For max simultaneous connections, I drop that back to the default 100000 Byte file for all pairs.

Ok and which results did you see when you hit max number of pairs, I tested with 300 pairs and more but I dont see something that I would say, ok I hit the max. number of pairs.

I'm interested what will fall out when I'm at the maximum number of pairs, is that non-responsive pairs or the router would die at the moment I hit the max number.

I saw something wonderful thing in this tp-link model, in fact in his qos settings you can bind range of ports and you could set the minimum guarantied bandwidth, if you use it cleverly you could make miracles because I could set minimal guarantied bandwidth for the port 80 for example 200kb/s and I could bind
higher ports that someone could use for torrents, nntp, ftp... into practically one policy with one minimum guarantied bandwidth, which means if someone start to browse he will momentary get the guarantied bandwidth allocation for http, and what's most important the ping would be fast as lightning, no matter if you get full bandwidth allocated to torrents, ftp or nntp.
When I tested dlink 625 I didn't get such http response when I had full bandwidth allocation. I'm sure that its because it didn't had minimal guarantied bandwidth policy.

I'm trying to get from the main distributor the Trendnet high class routers, I just want to compare tp-link with trendnet 633.
Today I watched their emulation of 633 on this link http://www.trendnet.com/emulators/TEW-633GR/Advanced/Traffic_Shaping.html
and I dont see advanced traffic shaping which means its on router to shape the traffic even if you set the priorities, because the creators were limited in the brain, like they dont understand that they need to fight one http connection in front of 200 torrents connections , or one http connection in front of 10 ftp connections or 1 http connection in front of 10 nntp connections, there is no way that router is clever enough to fight without the strict rules with the strict bandwidth numbers numbers.

The first time I saw minimal guarantied bandwidth in traffic shaping was with the oleg firmware on Asus router.

 

[thiggins]

If all pairs connect, pass traffic and properly terminate, the test passes.

I continue to add pairs until one or more pair fail the above criteria or I hit my IxChariot pair license limit, which is 200.

 

[ewenger]

RE: CHR0336 errors

Hi, I'm the Product Manager for IxChariot. I noticed the comments about the CHR0336 errors and I wanted to provide a little feedback. In my experience, the CHR0336 errors can frequently lead to errors in the performance calculations. In one example, I saw a 54 Mbps network reporting 67 Mbps of throughput when the max possible is around 29 Mbps.

It's always best to go in and pump up the file_size or transactions_per_record when you see that warning.

Best regards,
--Eric