Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: [802SEC] 802 reflector delays




I think the variable is something more subtle ...
perhaps other tasks (like backups or other maintenance
functions running on the server) ...

I monitored and responded to all of Bill's best messages,
with the sent timestamp and my received timestamp.

All of his test messages fell within about a 7 minute
or less latency, as I recall.  Some were delivered to
me within a minute or two.

16 hour delays indicate something else going on, IMHO.

Regards,
Carl


> -----Original Message-----
> From: Howard Frazier [mailto:millardo@dominetsystems.com]
> Sent: Saturday, March 08, 2003 3:16 PM
> To: Bill Quackenbush
> Cc: IEEE 802 SEC
> Subject: Re: [802SEC] 802 reflector delays
> 
> 
> 
> 
> One explanation for the extreme delay may be the relative length of
> the address lists for the efm and sec reflectors.  The efm 
> reflector has
> about 1300 subscribers (and billq's address is about the 
> 1000th entry),
> whereas the sec reflector is much, much smaller (51 entries,
> I believe).  I am not exactly sure how the resend subsystem works, but
> I will venture the guess that messages to each subscriber are 
> transmitted
> individually, so it may simply take a lot more time to walk 
> through all
> 1300 entries of the list.  We can test this by comparing the resend
> timestamps for messages sent to addresses early in a long 
> list to those
> sent to addresses farther down.
> 
> In the case of the first message sent by me at 05:22:49 GMT on March
> 6th, it appears that my pop3 server received the message from
> ruebert.ieee.org at 05:31:39 GMT on March 6th, for a delay of roughly
> nine minutes, whereas Bill's email server didn't receive it 
> for another
> 16 hours and 11 minutes.  My email address is the very first in the
> list on the stds-802-3-efm reflector, whereas Bill's is 
> number 1000 out
> of roughly 1300.
> 
> You could verify this theory by examining the received timestamps on
> the barrage of test messages that Bill recently sent, comparing the
> received timestamps of each of the members of the stds-802-sec
> list, and determining whether the received timestamps roughly
> correspond to the order in which the addresses appear on the list.
> I'll wager that (for example) kenneth.alonge@verizon.net consistently
> receives message with a later timestamp than tony@jeffree.co.uk.
> Any one want to take this wager? The bet is a beer in Dallas.
> 
> Proving once again that all animals are equal, but that some are more
> equal than others.
> 
> For those who are truly in need of a life, another 
> interesting exercise
> might be to estimate the time it takes for each copy of the message
> to be resent by ruebert.ieee.org, and to examine the delay variation,
> and investigate any possible correlation with time of day, 
> server load,
> and relaying delays across the internet.  The phase of the 
> moon may also
> prove to be a contributing factor. From this, we might be able to
> postulate an overall Quality of Service profile for sec 
> reflector messages
> from the perspective of each member of the list... (feel free 
> to shoot me at
> any time) I see an article for IEEE Communications in our future.
> 
> Howard
> 
> 
> Bill Quackenbush wrote:
> 
> >All,
> >
> >Two recent messages from Howard Frazier to the 802.3ah (EFM) 
> reflector
> >have has succeded where I failed.  Note the timestamps.
> >
> >-------------------
> >
> >Received: 
> >                          from ruebert.ieee.org ([140.98.193.10]) by
> >prserv.net (in1) with ESMTP id <2003030621473310100fu06ce>;
> >                          Thu, 6 Mar 2003 21:47:33 +0000
> >                 Received: 
> >                          (from daemon@localhost) by ruebert.ieee.org
> >(Switch-2.2.5/Switch-2.2.5) id h265QmH07762 for
> >                          stds-802-3-efm-resent; Thu, 6 Mar 2003
> >00:26:48 -0500 (EST)
> >               Message-ID: 
> >                          <3E66DB29.9040405@dominetsystems.com>
> >                     Date: 
> >                          Wed, 05 Mar 2003 21:22:49 -0800
> >                    From: 
> >                          Howard Frazier 
> <millardo@dominetsystems.com>
> >
> >--------------------
> >
> >Received: 
> >                          from ruebert.ieee.org ([140.98.193.10]) by
> >prserv.net (in3) with ESMTP id <20030306213849103034agn6e>;
> >                          Thu, 6 Mar 2003 21:38:50 +0000
> >                 Received: 
> >                          (from daemon@localhost) by ruebert.ieee.org
> >(Switch-2.2.5/Switch-2.2.5) id h266BVC27435 for
> >                          stds-802-3-efm-resent; Thu, 6 Mar 2003
> >01:11:31 -0500 (EST)
> >               Message-ID: 
> >                          <3E66E57D.6010102@dominetsystems.com>
> >                     Date: 
> >                          Wed, 05 Mar 2003 22:06:53 -0800
> >                    From: 
> >                          Howard Frazier 
> <millardo@dominetsystems.com>
> >
> >---------------------
> >
> >The first message was sent by Howard at 05:22:49 GMT, received by
> >ruebert.ieee.org at 05:26:48 GMT and was received from 
> ruebert.ieee.org
> >at 21:47:33 GNT, a delay by ruebert.ieee.org of 16 hours, 20 
> min and 45 sec.
> >
> >The second message was sent by Howard at 06:06:53 GMT, received by
> >ruebert.ieee.org at 06:11:31 GMT and was received from 
> ruebert.ieee.org
> >at 21:38:50 GNT, a delay by ruebert.ieee.org of 15 hours, 27 
> min and 19 sec.
> >
> >ruebert.ieee.org is the same reflector used by the 802 SEC alias.
> >
> >Thanks,
> >
> >wlq
> >
> >
> 
>