13:03 < Amanda_RIPENCC> Hi everyone. I'm Amanda from the RIPE NCC. We are about to start with the EIX session. If you have any questions or comments and want me to read it aloud to the speaker, please let me know. Clearly state your question/comment, your name and your affilation.
13:07 -!- prt-8676 [~prt [at] avalon [dot] prtsystems [dot] ltd [dot] uk] has joined #eix
13:08 -!- Chris_RIPENCC [~c100155e [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
13:08 < Amanda_RIPENCC> Christian Panigl, VIX is now presenting The Local Peering Scene âAll about Peering in Austria"
13:09 -!- ian-mat [~c348b361 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
13:12 -!- remco-eqix [~remcoeqix [at] dhcp-26-231 [dot] ripemtg [dot] ripe [dot] net] has joined #eix
13:13 -!- ripe290 [~a055263b [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
13:15 -!- ripe290 [~a055263b [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has quit [Quit: CGI:IRC (EOF)]
13:15 -!- ripe284 [~d4faefa2 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
13:23 -!- razvano [~razvano [at] dhcp-25-173 [dot] ripemtg [dot] ripe [dot] net] has joined #eix
13:23 -!- razvano is now known as razvan_ripencc
13:24 -!- razvan_ripencc is now known as razvan_RIPENCC
13:24 < Chris_RIPENCC> The next presentation is EIX Activity Updates â new IPv4 policy and switch wishlist by Andy Davidson, LONAP
13:31 < amd2-ripe> only for me the audio stream had a stop?
13:31 -!- ripe284 [~d4faefa2 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has quit [Quit: CGI:IRC (EOF)]
13:31 < ian-mat> stopped for me too
13:31 < amd2-ripe> ohwell, reload fixed it
13:31 < ian-mat> and video
13:32 < prt-8676> the video stream keeps stopping for me too.
13:32 < prt-8676> perhaps it doesn't like Andy :)
13:32 < amd2-ripe> hehe
13:33 -!- ian-mat [~c348b361 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has quit [Quit: CGI:IRC]
13:34 -!- ian-mat [~c348b361 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
13:34 -!- ivan_bev1 [~ivan_bev [at] dhcp-26-133 [dot] ripemtg [dot] ripe [dot] net] has joined #eix
13:35 < prt-8676> the video seems to be lagging badly behind realtime - the transcript is about two pages ahead of the video now.
13:36 -!- ian-mat [~c348b361 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has left #eix []
13:36 < Chris_RIPENCC> The next presentation is on Monitoring Platforms for Internet Exchange Points by Harald Michl
13:36 < amd2-ripe> prt-8676: for me it's spot on
13:36 < amd2-ripe> which is a little spooky with the transcriber knowing what people will say ;)
13:37 < prt-8676> hmm. a stop and restart has brought it back into line. How odd, has worked perfectly up to now.
13:38 -!- Chris_RIPENCC [~c100155e [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has left #eix []
13:39 -!- Chris_RIPENCC_ [~chrisb [at] dhcp-24-146 [dot] ripemtg [dot] ripe [dot] net] has joined #eix
13:43 < amd2-ripe> hmm.. ought to change nick during this
13:43 -!- amd2-ripe is now known as amd-DIX
13:47 -!- ian-mat [~c348b361 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
13:49 -!- dreamer [~7d1 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
13:55 < Chris_RIPENCC_> The next presentation is on the danger of proxy ARP in IX environment by Maksym Tulyuk, AMS-IX
13:55 < Chris_RIPENCC_> correction: Jumbo Frames in AMS-IX
13:55 < Chris_RIPENCC_> (but still by Maksym Tulyuk, AMS-IX)
13:56 -!- remco-eqix [~remcoeqix [at] dhcp-26-231 [dot] ripemtg [dot] ripe [dot] net] has quit [Quit: Colloquy for iPad - http://colloquy.mobi]
14:01 -!- lochii [lochii [at] singularity [dot] convergence [dot] cx] has joined #eix
14:01 < lochii> right, but, if you drop a frame, you lose a lot more
14:08 -!- remco-eqix [~remcoeqix@2001:67c:64:42:2a6a:baff:fe03:2053.] has joined #eix
14:12 -!- ripe161 [~7d1 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
14:12 < ripe161> ll
14:12 -!- ripe114 [~02 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
14:12 -!- Chris_RIPENCC [~chrisb [at] dhcp-24-146 [dot] ripemtg [dot] ripe [dot] net] has joined #eix
14:12 -!- Chris_RIPENCC_ [~chrisb [at] dhcp-24-146 [dot] ripemtg [dot] ripe [dot] net] has quit [Read error: Connection reset by peer]
14:14 -!- Chris_RIPENCC [~chrisb [at] dhcp-24-146 [dot] ripemtg [dot] ripe [dot] net] has quit [Remote host closed the connection]
14:14 -!- ytti [~504002f2 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has joined #eix
14:14 -!- Chris_RIPENCC [~chrisb@2001:67c:64:42:5a55:caff:fef6:b51b.] has joined #eix
14:14 < ytti> i'd peer with high mtu on amsix given chance
14:15 < ytti> we do so in netnod and ficix
14:15 < ytti> ficix ipv6 is 9k mtu only
14:16 < ytti> obviousle amsix would need netnod style both vlans
14:16 < ericAMSIX> At AMS-IX we frequently has complain about giant frames creating error on member interfaces :)
14:16 < ytti> you have arp scrubber, you could have 'mtu scrubber'
14:16 < ytti> pinging 1500B vlan interfaces with 1501B
14:16 < ytti> and 1499 DFbit
14:17 < ytti> and same for 9000B mtu
14:17 < ericAMSIX> Due to the slight increase in MTU size to cope with MPLS/VPLS
14:17 < ivan_bev1> At a previous commercial UK IXP (name escapes me at the moment), DSL providers used L2TP to speak to the provider-with-DSLAMs, and used jumbo frames for this. This was around 7 years ago, IIRC
14:17 < ytti> so you could proactively monitor member ports and report them when MTU is wrong
14:17 -!- ivan_bev1 is now known as ivan_bev
14:17 < lochii> 11:15]
14:17 < lochii> [11:15]
14:17 < lochii> as well as the large frames being dropped
14:17 < andyd> come to the mic lochii
14:17 < lochii> andyd: I'm not in the same room :(
14:18 < andyd> it's a great reason to announce your peering lan, and a great reason that 2011-05 is important. ;-)
14:18 < lochii> heh
14:18 < Chris_RIPENCC> lochii, i can read out statements from Jabber on the mic if you'd like
14:18 < lochii> I somehow knew you would say that
14:18 < lochii> Chris_RIPENCC: sure, you can, there are two issues
14:18 < Chris_RIPENCC> can you summarize them for me to read?
14:18 < lochii> the first being that if you are packing more information into these large frames, more is lost when they are dropped
14:19 < Chris_RIPENCC> sorry, can I get your real name as well?
14:19 < lochii> the second is that if your mtus are mismatched (i.e your peer does 9k and you dont') you have to issue a TOOBIG
14:19 < lochii> and also, what jen is saying, if you can't do PMTUD you are screwed
14:19 < lochii> david freedman
14:19 < ericAMSIX> I don't know if it a good practice to open the MTU size within AMS-IX and have one member pushing large MTU frames toward members with lower MTU configured :)
14:19 < ytti> if you can't do PMTUD you should send 576B only :)
14:20 -!- sherv [~Miranda [at] dhcp-27-14 [dot] ripemtg [dot] ripe [dot] net] has joined #eix
14:20 < ytti> or 1280B ipv6 :)
14:20 -!- GPF [~sebastian [at] gpf [dot] v6 [dot] karotte [dot] org] has joined #eix
14:21 < ytti> i see no problem have 4470B core and 9000B core
14:21 < ytti> and 9000B IXP MTU
14:22 < lochii> ytti: you use your outbound interface MTU
14:22 < lochii> you are thinking of BGP
14:22 * lochii thinking of just TCP
14:22 < Chris_RIPENCC> The next presentation is an APIX Update by Gaurab Raj Upadhaya
14:23 < ytti> BGP maximum is 4096B :>
14:23 < ytti> if your core is 4470B and partners core is 9000B, and IXP is 9000B. It'll just work
14:24 < ytti> all this PMTUD worry applies exactly the same for 1500B
14:25 -!- ripe161 [~7d1 [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has quit [Quit: CGI:IRC (Ping timeout)]
14:26 -!- remco-eqix [~remcoeqix@2001:67c:64:42:2a6a:baff:fe03:2053.] has quit [Ping timeout: 180 seconds]
14:28 -!- Amanda_RIPENCC [~agowland [at] dhcp-26-122 [dot] ripemtg [dot] ripe [dot] net] has quit [Quit: Amanda_RIPENCC]
14:28 < ericAMSIX> Regarding the IX environment, if we open up the MTU size within the peering VLAN, would it be safe for peer-2-peer traffic to self regulate MTU along the path ?
14:28 -!- sherv [~Miranda [at] dhcp-27-14 [dot] ripemtg [dot] ripe [dot] net] has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
14:29 -!- Chris_RIPENCC [~chrisb@2001:67c:64:42:5a55:caff:fef6:b51b.] has quit [Ping timeout: 180 seconds]
14:29 < dreamer> Possibly daft question: is there an option to download the transcript(s)?
14:30 < ytti> ericAMSIX: no, it's not safe to have arbitrary MTU in given VLAN
14:30 < ytti> you'd need separate VLAN for each MTU people need to use. Only practical solution is to agree to what is 'high mtu vlan', in my opnion
14:30 -!- ioc [~2e118f2d [at] rosie-arch [dot] ipv6 [dot] ripe [dot] net] has quit [Quit: CGI:IRC]
14:31 < ericAMSIX> That is why we brought up the question about separated JumboVLAN
14:31 < ytti> for my applications, i recognizise there are others too though, anything over 2000B is more than enough to me
14:31 < ytti> just as long as we can do consumer 1500B + overhead of ipsec/gre/pppoe whatnot
14:32 -!- matjaz [~matjaz [at] dhcp-26-180 [dot] ripemtg [dot] ripe [dot] net] has quit [Quit: matjaz]
14:32 < ytti> internet cores have long been higher mtu than consumer MTU, only thing stopping internet being, is IXPs
14:33 < ivan_bev> I guess email could also benefit from jumbos (server-server, rather than server-client), given the size of attachments these days
14:36 < ericAMSIX> So discussion will be held again @ AMS-IX general meeting (16th Nov 2011)
14:36 < ericAMSIX> About the MTU
14:36 -!- razvan_RIPENCC [~razvano [at] dhcp-25-173 [dot] ripemtg [dot] ripe [dot] net] has quit [Quit: razvan_RIPENCC]
14:37 < ericAMSIX> If it's members wish, we will work on the solution