[Cerowrt-devel] No legend for netperf-wrapper graphs

[Cerowrt-devel] No legend for netperf-wrapper graphs

[Fred Stratton]

[-- Attachment #1: Type: text/plain, Size: 228 bytes --]

If the legends are not explained, the graphs cannot be interpreted.

A list of working servers would help, also.

At present there appears to be only one in the observable universe.

fq_codel simplest.qos 6000 down 700 up htb.


[-- Attachment #2.1: Type: text/html, Size: 694 bytes --]

[-- Attachment #2.2: fq_simplest_htb_8000_700a.png --]
[-- Type: image/png, Size: 241826 bytes --]

Re: [Cerowrt-devel] No legend for netperf-wrapper graphs

[Dave Taht]

[-- Attachment #1.1: Type: text/plain, Size: 826 bytes --]

We have insufficient funding to maintain rrul servers everywhere they are
needed. There are servers in Germany and england if they are closer to you?
I can let you (or anyone else that asks) know the addresses privately... I
will probably add one in Japan soon and we continue to look for a hosting
service that will actively help....
On Aug 14, 2013 11:01 AM, "Fred Stratton" <fredstratton@imap.cc> wrote:

> If the legends are not explained, the graphs cannot be interpreted.
>
> A list of working servers would help, also.
>
> At present there appears to be only one in the observable universe.
>
> fq_codel simplest.qos 6000 down 700 up htb.
>
>
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>
>

[-- Attachment #1.2: Type: text/html, Size: 1401 bytes --]

[-- Attachment #2: fq_simplest_htb_8000_700a.png --]
[-- Type: image/png, Size: 241826 bytes --]

Re: [Cerowrt-devel] No legend for netperf-wrapper graphs

[Dave Taht]

[-- Attachment #1.1: Type: text/plain, Size: 1383 bytes --]

Probably the best explanation of the rrul test suite is in the talk I gave
at MIT. I and/or toke should try to put together a more comprehensive talk
just on it at this point

The markings tried in rrul are stuff that should exercise the WiFi 802.11e
queues and a few QoS systems if present. There are multiple other rrul
tests for things like RTT fairness and testing ipv4 and ipv6 at the same
time... Simpler ones for TCP and ping... Multiple graph types....
On Aug 14, 2013 11:07 AM, "Dave Taht" <dave.taht@gmail.com> wrote:

> We have insufficient funding to maintain rrul servers everywhere they are
> needed. There are servers in Germany and england if they are closer to you?
> I can let you (or anyone else that asks) know the addresses privately... I
> will probably add one in Japan soon and we continue to look for a hosting
> service that will actively help....
> On Aug 14, 2013 11:01 AM, "Fred Stratton" <fredstratton@imap.cc> wrote:
>
>> If the legends are not explained, the graphs cannot be interpreted.
>>
>> A list of working servers would help, also.
>>
>> At present there appears to be only one in the observable universe.
>>
>> fq_codel simplest.qos 6000 down 700 up htb.
>>
>>
>> _______________________________________________
>> Cerowrt-devel mailing list
>> Cerowrt-devel@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>>
>>

[-- Attachment #1.2: Type: text/html, Size: 2220 bytes --]

[-- Attachment #2: fq_simplest_htb_8000_700a.png --]
[-- Type: image/png, Size: 241826 bytes --]

Re: [Cerowrt-devel] No legend for netperf-wrapper graphs

[Fred Stratton]

[-- Attachment #1: Type: text/plain, Size: 1308 bytes --]

UK and German rrul server address would be helpful, on the basis that I shall be using them to test ADSL based configurations. For whatever reason I could not access the Yurt resident one.  (I thought that everyone who owned a Yurt maintained a breeding herd of goats or alpaca.)

You have given many talks. A link to the MIT talk would help.


On 14 Aug 2013, at 19:07, Dave Taht <dave.taht@gmail.com> wrote:

> We have insufficient funding to maintain rrul servers everywhere they are needed. There are servers in Germany and england if they are closer to you? I can let you (or anyone else that asks) know the addresses privately... I will probably add one in Japan soon and we continue to look for a hosting service that will actively help....
> 
> On Aug 14, 2013 11:01 AM, "Fred Stratton" <fredstratton@imap.cc> wrote:
> If the legends are not explained, the graphs cannot be interpreted.
> 
> A list of working servers would help, also.
> 
> At present there appears to be only one in the observable universe.
> 
> fq_codel simplest.qos 6000 down 700 up htb.
> 
> <fq_simplest_htb_8000_700a.png>
> 
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel
> 


[-- Attachment #2: Type: text/html, Size: 2217 bytes --]

Re: [Cerowrt-devel] No legend for netperf-wrapper graphs

[Fred Stratton]

[-- Attachment #1: Type: text/plain, Size: 1665 bytes --]

If it is the 'What is wrong with wifi talk, the I have found that.


On 14 Aug 2013, at 19:21, Fred Stratton <fredstratton@imap.cc> wrote:

> UK and German rrul server address would be helpful, on the basis that I shall be using them to test ADSL based configurations. For whatever reason I could not access the Yurt resident one.  (I thought that everyone who owned a Yurt maintained a breeding herd of goats or alpaca.)
> 
> You have given many talks. A link to the MIT talk would help.
> 
> 
> On 14 Aug 2013, at 19:07, Dave Taht <dave.taht@gmail.com> wrote:
> 
>> We have insufficient funding to maintain rrul servers everywhere they are needed. There are servers in Germany and england if they are closer to you? I can let you (or anyone else that asks) know the addresses privately... I will probably add one in Japan soon and we continue to look for a hosting service that will actively help....
>> 
>> On Aug 14, 2013 11:01 AM, "Fred Stratton" <fredstratton@imap.cc> wrote:
>> If the legends are not explained, the graphs cannot be interpreted.
>> 
>> A list of working servers would help, also.
>> 
>> At present there appears to be only one in the observable universe.
>> 
>> fq_codel simplest.qos 6000 down 700 up htb.
>> 
>> <fq_simplest_htb_8000_700a.png>
>> 
>> _______________________________________________
>> Cerowrt-devel mailing list
>> Cerowrt-devel@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>> 
> 
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel


[-- Attachment #2: Type: text/html, Size: 2943 bytes --]

Re: [Cerowrt-devel] No legend for netperf-wrapper graphs

[Fred Stratton]

[-- Attachment #1: Type: text/plain, Size: 2156 bytes --]

Nothing is mentioned in the MIT talk.

I f the servers are not available, so work cannot be duplicated, and the meaning of terms used is not revealed anywhere, the project ceases to be Open.


On 14 Aug 2013, at 20:50, Fred Stratton <fredstratton@imap.cc> wrote:

> If it is the 'What is wrong with wifi talk, the I have found that.
> 
> 
> On 14 Aug 2013, at 19:21, Fred Stratton <fredstratton@imap.cc> wrote:
> 
>> UK and German rrul server address would be helpful, on the basis that I shall be using them to test ADSL based configurations. For whatever reason I could not access the Yurt resident one.  (I thought that everyone who owned a Yurt maintained a breeding herd of goats or alpaca.)
>> 
>> You have given many talks. A link to the MIT talk would help.
>> 
>> 
>> On 14 Aug 2013, at 19:07, Dave Taht <dave.taht@gmail.com> wrote:
>> 
>>> We have insufficient funding to maintain rrul servers everywhere they are needed. There are servers in Germany and england if they are closer to you? I can let you (or anyone else that asks) know the addresses privately... I will probably add one in Japan soon and we continue to look for a hosting service that will actively help....
>>> 
>>> On Aug 14, 2013 11:01 AM, "Fred Stratton" <fredstratton@imap.cc> wrote:
>>> If the legends are not explained, the graphs cannot be interpreted.
>>> 
>>> A list of working servers would help, also.
>>> 
>>> At present there appears to be only one in the observable universe.
>>> 
>>> fq_codel simplest.qos 6000 down 700 up htb.
>>> 
>>> <fq_simplest_htb_8000_700a.png>
>>> 
>>> _______________________________________________
>>> Cerowrt-devel mailing list
>>> Cerowrt-devel@lists.bufferbloat.net
>>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>>> 
>> 
>> _______________________________________________
>> Cerowrt-devel mailing list
>> Cerowrt-devel@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
> 
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel


[-- Attachment #2: Type: text/html, Size: 3884 bytes --]

Re: [Cerowrt-devel] No legend for netperf-wrapper graphs

[Sebastian Moeller]

Hi there,

I am a bit grumpy right now, since it seems as if my router just died on me (trying an "emergency" flash to 3.10.6).

On Aug 14, 2013, at 23:05 , Fred Stratton <fredstratton@imap.cc> wrote:

> Nothing is mentioned in the MIT talk.
> 
> I f the servers are not available, so work cannot be duplicated,

	The buffer bloat effort is mostly driven by enthusiasm not monetary gains. So as I understand it netperf servers are run by volunteers out of their own pocket. Given that a number of ISPs have maximum transfer volumes by months I think it actually is amazing that there are some server available at all. I think that no one would complain and rather support it if you would volunteer to set up a netperf server to remedy your observed lack of openness, allowing others to test against your router. :)

> and the meaning of terms used is not revealed anywhere, the project ceases to be Open.

	Now have a look at https://github.com/tohojo/netperf-wrapper/blob/master/tests/rrul.conf, and http://en.wikipedia.org/wiki/Differentiated_services you can get a pretty good overview of what the diffent line colors per plot actually mean. My take is for the ADSL encapsulation tests all that matters is the average and the fact that is is averaged from 4 flows, so mentally multiplying the y-axis by 4 gives an idea about the total bandwidth used. The interesting stuff happens in the ping RTT plots, which you want to be as flat as possible with preferably no big increase once the TCP loads start.


Best Regards
	Sebastian


> 
> 
> On 14 Aug 2013, at 20:50, Fred Stratton <fredstratton@imap.cc> wrote:
> 
>> If it is the 'What is wrong with wifi talk, the I have found that.
>> 
>> 
>> On 14 Aug 2013, at 19:21, Fred Stratton <fredstratton@imap.cc> wrote:
>> 
>>> UK and German rrul server address would be helpful, on the basis that I shall be using them to test ADSL based configurations. For whatever reason I could not access the Yurt resident one.  (I thought that everyone who owned a Yurt maintained a breeding herd of goats or alpaca.)
>>> 
>>> You have given many talks. A link to the MIT talk would help.
>>> 
>>> 
>>> On 14 Aug 2013, at 19:07, Dave Taht <dave.taht@gmail.com> wrote:
>>> 
>>>> We have insufficient funding to maintain rrul servers everywhere they are needed. There are servers in Germany and england if they are closer to you? I can let you (or anyone else that asks) know the addresses privately... I will probably add one in Japan soon and we continue to look for a hosting service that will actively help....
>>>> 
>>>> On Aug 14, 2013 11:01 AM, "Fred Stratton" <fredstratton@imap.cc> wrote:
>>>> If the legends are not explained, the graphs cannot be interpreted.
>>>> 
>>>> A list of working servers would help, also.
>>>> 
>>>> At present there appears to be only one in the observable universe.
>>>> 
>>>> fq_codel simplest.qos 6000 down 700 up htb.
>>>> 
>>>> <fq_simplest_htb_8000_700a.png>
>>>> 
>>>> _______________________________________________
>>>> Cerowrt-devel mailing list
>>>> Cerowrt-devel@lists.bufferbloat.net
>>>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>>>> 
>>> 
>>> _______________________________________________
>>> Cerowrt-devel mailing list
>>> Cerowrt-devel@lists.bufferbloat.net
>>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>> 
>> _______________________________________________
>> Cerowrt-devel mailing list
>> Cerowrt-devel@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
> 
> _______________________________________________
> Cerowrt-devel mailing list
> Cerowrt-devel@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cerowrt-devel

Re: [Cerowrt-devel] No legend for netperf-wrapper graphs

[Fred Stratton]

If it is confirmed that the servers are run by volunteers, I would be happy to contribute. I note their silence over past months. Running a server here is of little help.

Again, the Comcast funding involvement is unexplained. If they are funding the open vehicle, then they could pay.

I appreciate the links, but the point stands. Better I make the criticism than others less favourably disposed.

As for your router, it should be possible to restore it via a serial-usb cable, as I am sure you know. Have you tried to bring up telnet, like OpenWRT? That mechanism may be present, but I do not know what the default IP address would be.

I have posted td-stab and hub rrul graphs, but post is large and has gone for moderation.




On 14 Aug 2013, at 22:27, Sebastian Moeller <moeller0@gmx.de> wrote:

> Hi there,
> 
> I am a bit grumpy right now, since it seems as if my router just died on me (trying an "emergency" flash to 3.10.6).
> 
> On Aug 14, 2013, at 23:05 , Fred Stratton <fredstratton@imap.cc> wrote:
> 
>> Nothing is mentioned in the MIT talk.
>> 
>> I f the servers are not available, so work cannot be duplicated,
> 
> 	The buffer bloat effort is mostly driven by enthusiasm not monetary gains. So as I understand it netperf servers are run by volunteers out of their own pocket. Given that a number of ISPs have maximum transfer volumes by months I think it actually is amazing that there are some server available at all. I think that no one would complain and rather support it if you would volunteer to set up a netperf server to remedy your observed lack of openness, allowing others to test against your router. :)
> 
>> and the meaning of terms used is not revealed anywhere, the project ceases to be Open.
> 
> 	Now have a look at https://github.com/tohojo/netperf-wrapper/blob/master/tests/rrul.conf, and http://en.wikipedia.org/wiki/Differentiated_services you can get a pretty good overview of what the diffent line colors per plot actually mean. My take is for the ADSL encapsulation tests all that matters is the average and the fact that is is averaged from 4 flows, so mentally multiplying the y-axis by 4 gives an idea about the total bandwidth used. The interesting stuff happens in the ping RTT plots, which you want to be as flat as possible with preferably no big increase once the TCP loads start.
> 
> 
> Best Regards
> 	Sebastian
> 
> 
>> 
>> 
>> On 14 Aug 2013, at 20:50, Fred Stratton <fredstratton@imap.cc> wrote:
>> 
>>> If it is the 'What is wrong with wifi talk, the I have found that.
>>> 
>>> 
>>> On 14 Aug 2013, at 19:21, Fred Stratton <fredstratton@imap.cc> wrote:
>>> 
>>>> UK and German rrul server address would be helpful, on the basis that I shall be using them to test ADSL based configurations. For whatever reason I could not access the Yurt resident one.  (I thought that everyone who owned a Yurt maintained a breeding herd of goats or alpaca.)
>>>> 
>>>> You have given many talks. A link to the MIT talk would help.
>>>> 
>>>> 
>>>> On 14 Aug 2013, at 19:07, Dave Taht <dave.taht@gmail.com> wrote:
>>>> 
>>>>> We have insufficient funding to maintain rrul servers everywhere they are needed. There are servers in Germany and england if they are closer to you? I can let you (or anyone else that asks) know the addresses privately... I will probably add one in Japan soon and we continue to look for a hosting service that will actively help....
>>>>> 
>>>>> On Aug 14, 2013 11:01 AM, "Fred Stratton" <fredstratton@imap.cc> wrote:
>>>>> If the legends are not explained, the graphs cannot be interpreted.
>>>>> 
>>>>> A list of working servers would help, also.
>>>>> 
>>>>> At present there appears to be only one in the observable universe.
>>>>> 
>>>>> fq_codel simplest.qos 6000 down 700 up htb.
>>>>> 
>>>>> <fq_simplest_htb_8000_700a.png>
>>>>> 
>>>>> _______________________________________________
>>>>> Cerowrt-devel mailing list
>>>>> Cerowrt-devel@lists.bufferbloat.net
>>>>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>>>>> 
>>>> 
>>>> _______________________________________________
>>>> Cerowrt-devel mailing list
>>>> Cerowrt-devel@lists.bufferbloat.net
>>>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>>> 
>>> _______________________________________________
>>> Cerowrt-devel mailing list
>>> Cerowrt-devel@lists.bufferbloat.net
>>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>> 
>> _______________________________________________
>> Cerowrt-devel mailing list
>> Cerowrt-devel@lists.bufferbloat.net
>> https://lists.bufferbloat.net/listinfo/cerowrt-devel
>