Re: [tcpm] Fwd: Re: Data on 'Nonce' and 'Broken' responses to AccECN SYN?

[Bob Briscoe <[e-mail]>]

Michael,

I agree. Also, even tho 'broken' is not currently any use, it might be 
one day.

So we should reserve both the nonce and broken patterns on the SYN-ACK 
(in response to 111 on the SYN) for future AccECN use. They have to be 
solely for AccECN use, cos the SYN is still saying "I support AccECN".


We can add a RESVD tag in table 2, and explain in the notes underneath.


Bob

On 18/07/18 09:24, Scharf, Michael (Nokia - DE/Stuttgart) wrote:
>
> Interesting. So what prevents us from reserving the nonce pattern for 
> future use?
>
> Michael
>
> *From:*tcpm [mailto:[e-mail]] *On Behalf Of *Bob Briscoe
> *Sent:* Wednesday, July 18, 2018 1:41 PM
> *To:* Michael Scharf <[e-mail]>
> *Cc:* ANNA MARIA MANDALARI <[e-mail]>; tcpm IETF list 
> <[e-mail]>
> *Subject:* [tcpm] Fwd: Re: Data on 'Nonce' and 'Broken' responses to 
> AccECN SYN?
>
> Michael,
>
>
> Below is data from 410,803 of the Alexa top 500k web server that 
> confirms what I said in the presentation about space for future 
> evolution of AccECN:
>
>   * the nonce pattern on the SYN-ACK could be reused now (0.0007%)
>   * the broken reflection pattern is still far too prevalent to re-use
>     it (0.35%).
>
> If anyone wants the list of broken servers that Anna attached for me, 
> pls ask.
> The whole dataset is also available via: 
> http://www.it.uc3m.es/amandala/ecn++/ 
> <http://www.it.uc3m.es/amandala/ecn++/>
>
> @Anna, thanks v much for responding so quickly - it was me that hadn't 
> read your email in time for the talk.
>
>
>
> Bob
>
> -------- Forwarded Message --------
>
> *Subject: *
>
> 	
>
> Re: Data on 'Nonce' and 'Broken' responses to AccECN SYN?
>
> *Date: *
>
> 	
>
> Tue, 17 Jul 2018 15:42:10 +0200
>
> *From: *
>
> 	
>
> ANNA MARIA MANDALARI <[e-mail]> <mailto:[e-mail]>
>
> *To: *
>
> 	
>
> Bob Briscoe <[e-mail]> <mailto:[e-mail]>
>
> Hi Bob,
>
> I had a look at the data and I found *1,438 servers *over the top 
> 410,803 Alexa (0.35%) that reply SYN/ACK+111 to a SYN+111 (attached 
> the list).
>
> Only *3 servers *(0.0007%) reply SYN/ACK+101 to a SYN+111:
>
> test_synack;200.12.171.53;80;ect;0;flags;338
> test_synack;60.28.220.134;80;ect;0;flags;338
> test_synack;200.12.171.52;80;ect;0;flags;338
>
> Let me know if I can help you with something else!
>
> 2018-07-17 13:26 GMT+02:00 Bob Briscoe <[e-mail] 
> <mailto:[e-mail]>>:
>
>     Anna,
>
>     Could you do me a favour and look up how many servers responded to
>     an AccECN (111) SYN with a SYN/ACK carrying respectively 'Nonce'
>     (101) or 'Broken' (111)?
>
>     And also the total number of tests sent with an AccECN SYN, so I
>     can give proportions in the AccECN draft.
>
>     Cheers
>
>
>
>     Bob (too lazy to look at the data myself)
>
>     -- 
>     ________________________________________________________________
>     Bob Briscoe http://bobbriscoe.net/
>
>
>
>
> -- 
>
> ANNA MARIA MANDALARI
> Universidad Carlos III de Madrid
>

-- 
________________________________________________________________
Bob Briscoe                               http://bobbriscoe.net/