IETF
bmwg
bmwg@jabber.ietf.org
Thursday, 28 July 2011< ^ >
Room Configuration

GMT+0
[16:15:39] rajiv.family joins the room
[16:23:08] rajiv.family leaves the room: offline
[16:52:48] acmacm joins the room
[16:53:04] <acmacm> we will start at 13:00
[16:57:36] miyano joins the room
[16:58:03] rajiv.family joins the room
[16:58:24] rajiv.family leaves the room: offline
[17:01:01] Rajiv joins the room
[17:01:20] bhavani@cisco.com joins the room
[17:02:48] paul.aitken joins the room
[17:04:11] Chris joins the room
[17:04:42] Mike Hamilton joins the room
[17:05:10] Will Ivancic joins the room
[17:05:42] vijay.gurbani joins the room
[17:06:35] <Rajiv> Hi Al, can you please be a little louder or close to the mic
[17:07:07] <Will Ivancic> fowarded it on
[17:07:29] <Will Ivancic> is that better?
[17:07:40] <Rajiv> yes a bit
[17:08:26] <vijay.gurbani> Where are we in the agenda?
[17:08:33] <Will Ivancic> covering items for those present first as some are not back from lunch yet
[17:08:46] <Will Ivancic> bmwg activity report
[17:08:56] <vijay.gurbani> Thanks.
[17:10:16] Sarah Banks joins the room
[17:14:01] vijay.gurbani leaves the room
[17:16:23] <paul.aitken> OK, will do.
[17:16:59] vijay.gurbani joins the room
[17:17:54] <Chris> Brian Trammell will review as well.
[17:18:35] Ilya Varlashkin joins the room
[17:18:43] <Ilya Varlashkin> hello
[17:19:08] <Ilya Varlashkin> which presentation?
[17:19:17] <Sarah Banks> Sip Benchmarking
[17:19:27] <Ilya Varlashkin> 10x
[17:26:17] paul.aitken leaves the room
[17:26:58] <Sarah Banks> Moving on to Content aware network devices
[17:28:59] vijay.gurbani leaves the room
[17:29:34] <Ilya Varlashkin> what will happen when you benchmark WAN accellerators? are they qualified for this draft?
[17:30:18] <Ilya Varlashkin> they reduce traffic
[17:30:38] <Ilya Varlashkin> i.e. always less out than in
[17:30:45] <Ilya Varlashkin> me
[17:30:55] <Ilya Varlashkin> I did
[17:30:59] <Ilya Varlashkin> yes
[17:31:06] <Sarah Banks> Ilya can you review the doc and give me and Mike feedback please?
[17:31:14] <Ilya Varlashkin> yes
[17:31:15] <Sarah Banks> Mike and I. You get the idea :)
[17:31:17] <Sarah Banks> thank you
[17:31:21] <Ilya Varlashkin> :-D
[17:32:50] <Ilya Varlashkin> oh no, I hope jabber.org doesn't act odd today again
[17:33:52] <Rajiv> for me it's the audio today
[17:34:25] <Rajiv> very difficult to understand
[17:34:42] <Ilya Varlashkin> audio is fine, though have to crank volume
[17:34:49] <Rajiv> ;)
[17:34:57] <Rajiv> it is max for me
[17:40:23] Rajiv leaves the room
[17:40:57] rjs joins the room
[17:40:59] rjs is now known as rjsPad
[17:41:09] <Ilya Varlashkin> hey rob
[17:41:29] <rjsPad> HI Ilya.
[17:42:16] <rjsPad> Thanks for bringing this one to my attention :-)
[17:42:34] <Ilya Varlashkin> we're still missing our presenter as none of the authors is onsite
[17:43:30] <Ilya Varlashkin> why not do like: test without malformed for perf. then measure WHEN will box fail if malformed present?
[17:43:38] <Ilya Varlashkin> Sarah, can you pass that?
[17:44:11] <Sarah Banks> Chris is doing a great job in the voice of Ilya here :)
[17:44:27] <Ilya Varlashkin> ah :-)
[17:44:55] <Chris> About as much as I can useful contribute. ;)
[17:45:15] <Ilya Varlashkin> I mean like with rfc2544 - we don't send rogue frames/packets and try to determine perf under that condition
[17:45:41] <Ilya Varlashkin> o/ I did
[17:45:56] <Chris> should I mic that?
[17:45:59] <Sarah Banks> no
[17:46:00] <Sarah Banks> :)
[17:46:03] <Ilya Varlashkin> for rfc2544? yes
[17:46:28] <Chris> 1 sec
[17:47:06] <Sarah Banks> Ilya i meant the "i read it" part :)
[17:47:13] <Ilya Varlashkin> LOL
[17:48:30] <Ilya Varlashkin> exactly
[17:52:04] <Ilya Varlashkin> greg there?
[17:52:12] <Ilya Varlashkin> great
[17:52:35] <Sarah Banks> Moving onto BGP Data plane benchmarkin
[17:52:36] rajiv.family_gmail.com joins the room
[17:52:44] <Sarah Banks> Greg says he'll do his best to twist Ilya's words
[17:52:49] <Sarah Banks> err i mean he'll try NOT to
[17:52:51] <Sarah Banks> :)
[17:52:52] <rajiv.family_gmail.com> I am back
[17:53:04] <Ilya Varlashkin> rajiv just in time :-)
[17:55:13] rajiv.family_gmail.com leaves the room: Logged Out
[17:55:24] rajiv.family_gmail.com joins the room
[17:56:16] <Ilya Varlashkin> splitting would speed up work
[17:58:07] <Ilya Varlashkin> i.e. assuming devices behind DUT are dumb
[17:59:00] <Ilya Varlashkin> new draft is sort of independent from protocol being carried in data packets
[17:59:07] <Sarah Banks> don't you get into the sticky land then, trying to define what "dumb" is?
[17:59:38] <Ilya Varlashkin> that we assume they didn't hear that failure occured
[17:59:57] <Ilya Varlashkin> they keep sending to DUT so DUT alone can restore traffic flow
[18:00:40] <Ilya Varlashkin> same steps, just different failure scenarios, same topology
[18:00:50] <Ilya Varlashkin> al saw part of new draft :-)
[18:00:59] <Ilya Varlashkin> stop
[18:01:02] <Ilya Varlashkin> I elaborate
[18:01:28] <Ilya Varlashkin> new draft assumes devices on left-hand side on slide 5 do not know about failure
[18:01:33] <Ilya Varlashkin> and keep sending packets to dut
[18:01:48] <Ilya Varlashkin> so DUT alone is responsible to restre traffic ASAP
[18:02:01] lochii joins the room
[18:02:27] <Sarah Banks> greg commented that that's a special case ilya
[18:02:32] <Ilya Varlashkin> 8
[18:02:40] <Sarah Banks> is that what you intend the potential new draft to be based on.. a special corner case?
[18:03:01] <Ilya Varlashkin> no, that's not corner case. in fact it's The Most Generic case
[18:03:04] <bhavani@cisco.com> its not a corner case
[18:03:06] <rjsPad> Would be interested to see what you are defining in this background topological noise Ilya - had significant problems with defining this before.
[18:03:20] <Ilya Varlashkin> rjs: moment
[18:03:56] <bhavani@cisco.com> link & node failure scenarios when we have local restoration capability available
[18:04:12] <Ilya Varlashkin> obviously new draft tests not much BGP but local-repair
[18:04:29] <rajiv.family_gmail.com> independent of the protocols enabled
[18:04:37] <Ilya Varlashkin> I'll be in Taiwan (got ticket already) so I'll explain more next meeting
[18:04:51] <Chris> (you guys are up on the screen with the slides now)
[18:04:55] <Sarah Banks> you could help expedite on the list too, ilya
[18:04:59] <Sarah Banks> email list, i mean
[18:05:06] <rajiv.family_gmail.com> we plan on submitting both before the next meeting
[18:05:23] <Ilya Varlashkin> the new draft is largely ready, needs some editorial before publishing
[18:05:56] <Ilya Varlashkin> :-)
[18:06:05] <Ilya Varlashkin> I copied my other draft :-)
[18:06:46] <Ilya Varlashkin> can group accommodate new direction?
[18:06:54] <Ilya Varlashkin> i.e. is it good thing or bad thing?
[18:07:12] <rajiv.family_gmail.com> we present the two documents, and then the group can decide if they can be merged
[18:07:50] <rajiv.family_gmail.com> ok
[18:08:25] <Ilya Varlashkin> thanks to greg
[18:08:30] <rajiv.family_gmail.com> thank you Greg
[18:08:39] <bhavani@cisco.com> thx Greg
[18:10:08] <Ilya Varlashkin> we had one or two discussions
[18:10:53] rajiv.family_gmail.com leaves the room
[18:11:48] <Ilya Varlashkin> yes
[18:15:40] <Sarah Banks> Moving on from the work proposal slide, onto the IMIX Genome
[18:15:50] <Sarah Banks> Al presenting.
[18:17:48] <Ilya Varlashkin> question I forgot to ask earlier - if mix is LONG
[18:17:57] <Chris> yes?
[18:18:07] <Ilya Varlashkin> could this somehow be shortened?
[18:18:41] <Ilya Varlashkin> I mean how anyone is going to document traffic mix that fills only 1 second worth of 10gigabit pipe?
[18:19:01] <Sarah Banks> that's what the genome would do, right?
[18:19:04] <Ilya Varlashkin> I like sequence more
[18:19:08] <Sarah Banks> it's irrespective of how LONG it's in the pipe for
[18:19:13] <Sarah Banks> time isn't a factor, i thought
[18:19:16] <lochii> its just a sequence
[18:19:23] <lochii> you repeat it at will
[18:19:24] <Ilya Varlashkin> yes, very long sequence
[18:19:47] <Ilya Varlashkin> nono, but allow some numbers if more than one packet of same type
[18:20:03] <Ilya Varlashkin> like in chemical formulae
[18:20:19] <Ilya Varlashkin> yes
[18:20:45] <Ilya Varlashkin> it's much older than internet :-)
[18:20:49] <Ilya Varlashkin> I'll get on the list
[18:21:51] <Chris> something simple like postfix repeats: (aabedfecabecd)3 repeats the sequence 3 times, right?
[18:21:58] <Ilya Varlashkin> yes
[18:22:01] <lochii> rle works well
[18:22:42] <Ilya Varlashkin> o/
[18:22:49] <lochii> what about if you are dealing with really large sizes, and you use mtu
[18:22:57] <lochii> you then only have one really large size
[18:23:03] <Ilya Varlashkin> hmhmhmhmhmhmhmhmhm
[18:23:04] <lochii> no option for like 1/2 mtu or 1/4 mtu?
[18:23:25] <Sarah Banks> ilya i can't tell if you were for or against concensus but Ron called it anyway :)
[18:23:27] <Ilya Varlashkin> lochii: mail to the list
[18:23:31] <lochii> aye
[18:25:19] <Ilya Varlashkin> but people do measure LIVE networks and there are commercial products that stipulate it
[18:26:04] <Ilya Varlashkin> yes specifically
[18:26:13] <Ilya Varlashkin> and they probably do sometimes
[18:27:21] <Ilya Varlashkin> we can say "this is not intended use" but we can't enforce that
[18:27:46] <Ilya Varlashkin> and just imagine people start applying convergence benchmarking to live networks
[18:28:35] <Ilya Varlashkin> may be stay in big friendly letters "DON"T USE BMWG methods on live net - you will break the internet"
[18:28:44] <Ilya Varlashkin> and not only 2544
[18:29:01] <Ilya Varlashkin> s/stay/say/
[18:30:34] <Ilya Varlashkin> publish BCP "1001 way to break your network when you apply BMWG"
[18:34:23] <Ilya Varlashkin> Q: what exactly goal? to discourage people doing silly things? or to encourage every author to include section in the draft?
[18:34:32] <Ilya Varlashkin> IGP data plane
[18:35:21] <Ilya Varlashkin> o/
[18:36:04] <Sarah Banks> Ilya you're going to owe Chris a beverage after this meeting i suspect :)
[18:36:08] <lochii> heh
[18:36:13] <Ilya Varlashkin> so then BCP for whole BMWG not just 2544
[18:36:46] <Ilya Varlashkin> I'll be delighted to buy Chris cold beer in Taiwan
[18:36:53] <lochii> surely, it is a wg scope issue
[18:36:58] <lochii> I dont think anything is in scope for live networks
[18:37:28] <Ilya Varlashkin> accept
[18:37:33] <lochii> accept what?
[18:37:38] <Ilya Varlashkin> accept draft :-)
[18:37:51] <Ilya Varlashkin> lochii: you are not listenin to audio?
[18:37:56] <lochii> I'm in the room
[18:37:59] <Ilya Varlashkin> can I give brand names here?
[18:38:00] <Sarah Banks> it achieved consensus
[18:38:07] <lochii> you just said "accept" when we were humming for reject
[18:38:22] <lochii> (or perhaps audio lag?)
[18:38:23] <Ilya Varlashkin> lochii: audio lags
[18:38:26] <lochii> yeah
[18:38:29] <Chris> (I didn't voice that part…) :)
[18:38:38] <Ilya Varlashkin> Sarah: is saying brand names ok here?
[18:38:49] <Sarah Banks> i … don't know
[18:39:02] <Ilya Varlashkin> ask Al? :-)
[18:39:19] <Ilya Varlashkin> because I have names of few offenders of 2544
[18:39:31] <Ilya Varlashkin> i.e. those who claim testing 2544 on live nets
[18:39:32] <Sarah Banks> I think Al might have moved on
[18:39:40] <Sarah Banks> but I can ask Chris to repeat
[18:39:50] <Ilya Varlashkin> at the session end or ask chris
[18:39:54] <Sarah Banks> ok
[18:40:46] <Ilya Varlashkin> ok :-)
[18:42:15] lochii leaves the room
[18:42:18] Mike Hamilton leaves the room
[18:42:20] Chris leaves the room
[18:42:26] <Ilya Varlashkin> cheers
[18:42:38] <Sarah Banks> bye folks
[18:42:41] <Sarah Banks> thanks ilya, for the feedback
[18:42:59] Sarah Banks leaves the room
[18:44:17] Will Ivancic leaves the room
[18:44:48] bhavani@cisco.com leaves the room
[18:48:20] acmacm leaves the room
[18:58:32] Ilya Varlashkin leaves the room
[19:01:26] miyano leaves the room
[19:07:44] rjsPad leaves the room
[19:19:19] miyano joins the room
[19:27:44] miyano leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!