- ipv6-wg - mailman.ripe.net

[ipv6-wg@ripe.net] Agenda for IPv6 wg RIPE 44
by David Kessens 21 Jan '03

21 Jan '03

Hi, Below follows the agenda for the ipv6 wg for RIPE44. I would like to draw your attention to the agendas of the other wg session too since many have ipv6 agenda items too this time. As an experiment, in order to help to achieve the goal of less scheduling conflicts, I offered to shorten the ipv6 wg session to one 2 hour slot to help create more time in the schedule. As a result, the agenda is fairly full, so I won't have a lot of flexibility for last minute items. Thanks & I hope to see you all in Amsterdam, David K. --- Agenda for the IPv6 Working Group Meeting RIPE44 A. Administrative stuff - appointment of scribe - agenda bashing (David Kessens) B. IPv6 version of TTM (Henk Uijterwaal) C. IPv6 whois proxy (Shane Kerr) D. IPv6 state of art in the italian educational and research community. (Gabriella Paolini) E. DVTS (high bandwidth digital video) over 6PE (IPv6 over MPLS) over UTI over IPv4 (Erik-Jan Bos) F. Input for/from the DNS wg: Action point 42.1: Investigate the CNAME and other solutions for v6-reverse delegation (RIPE NCC) IPv6 capable RR DNS servers - what is needed, what are the issues ?!? (RIPE NCC) Action point 42.2: Give an overview of 6-to-4 reverse delegation issues at RIPE 43 (RIPE NCC) G. Global IPv6 routing table status (Gert Doering) H. Report(s) about *actual* v6 traffic volume as compared to v4? *what's real* out there, not what's on powerpoint? http://www.ipv6.aorta.net (Roger Jorgenson) http://www.nlnetlabs.nl/ipv6/measurements/index.en.html (Ronald van der Pol) (other reports welcome!) I. Workable approaches for ipv6 multihoming in the absence of a better solution than the one we use in the ipv4 Internet (Kurt Erik Lindqvist, Iljitsch van Beijnum) J. Fragmentation of the ipv6 address space from the top down by allocating /23's to RIRs (brief discussion) K. Developments/initiatives regarding IPv6 in the RIPE region and beyond (input from the audience) L. input to the RIPE NCC Activity Plan & results of KPMG survey (input from the audience) Z. AOB ---

1 0

[ipv6-wg@ripe.net] We need you..
by Timo Dreger 18 Jan '03

18 Jan '03

Hello, now the project SIX-Net.org is online. But our content is very small. We need a Server, pls look at the page. And if you like to sponsor us a server, mail me! SIX-Net offers free IPv6 services and we need resources. The services are as example: irc-bouncer, ircd, AAAA-Records, v6-Hosting, timeserver, ... and all for free ;-) At the time the page is only available in german. Thanks.. -- Timo Dreger eMail: timo.dreger(a)sobdog.de -- My Private Page: My IPv6-Project: v4: http://www.sobdog.de v4: http://www.six-net.org v6: http://ipv6.sobdog.de v6: http://ipv6.six-net.org --

1 0

[ipv6-wg@ripe.net] Minutes of ipv6 wg session RIPE43
by David Kessens 11 Jan '03

11 Jan '03

Hi, No comments were received on the draft minutes as posted to this list in december 2002. Therefore, the minutes as attached below can now be considered the final version. David K. --- Draft minutes (v1) ipv6 wg RIPE 43 Rhodes, Greece 11 September 2002 A. Administrative stuff ======================= Agenda is available at: http://www.kessens.com/~david/presentations/ The scribe was appointed, and the agenda was agreed upon. B. Status of the 6bone (David Kessens (DK)) ====================== Presentation is available at: http://www.kessens.com/~david/presentations/ Per 20020911, there are now 1187 ipv6 sites in 58 different countries on the ipv6 Internet. C. IPv6 version of TTM (Henk Uijterwaal (HU)) ====================== Presentation is available at: http://www.ripe.net/ripe/meetings/archive/ripe-43/ presentations/ripe43-ipv6-ttm/ Wilfried Woeber (WW): With 6net hat - is possible to make preliminary measurements available before the end of the year? DK: How is this service supported from the commercial point of view? Right now, you have to pay for the test-box. HU: You still have to buy a test box, and we operate it for you. Service fee is currently 3000 EUR. Q: Do you get v4 measurements with that, too? A: In principle, yes. WW: Technical question: the ntp protocol is used to encode v4 from clock-host partners to prevent forming loops. Anyone here knows more? HU: forward me the emails from the discussion. DK: I like this new service. This can be a reason for me to order such a box, because v6 production is important for us, and I want to be able to measure the performance. D. Action point 42.1: ===================== Investigate the CNAME and other solutions for v6-reverse delegation E. IPv6 capable RR DNS servers ============================== What is needed, what are the issues ?!? (RIPE NCC) F. Action point 42.2: ===================== Give an overview of 6-to-4 reverse delegation issues at RIPE 43 (RIPE NCC) Andrei Robachevski, RIPE NCC covered topic D, E & F. In regards to D & E, a prototype was made; setups how to support them was tested; no final decision yet. Status: in progress. It will be accomplished before the next ripe meeting Mohsen Souissi (MS): Are you aware of the technical notes, published last week, about DNAME, by the IETF committee? This is only meant for the smooth transition from .int -> .arpa Things are moving very very slowly. There is no dns server for ip6.arpa. "ns.ripe.net" is not ipv6-capable. I have already suggested several times to IANA that AFNIC server is available to be a secondary for ip6.arpa, but I received no answer... DK: There is an agenda point in the AOB about the Activity Plan of the RIPE NCC. We need to add this in there. My suggestion is that the RIPE NCC members make it clear to the RIPE NCC and the RIPE NCC Board that we want RIPE NCC to give some real attention to this. G. Global IPv6 routing table status (Gert Doering (GD)) =================================== Presentation is available at: http://www.ripe.net/ripe/meetings/archive/ripe-43/ presentations/ripe43-ipv6-table/ Q: Do you see people changing over, or announcing both? A: It was possible to see everything, all combinations. Q: Are the filtering problems still there? A: Not really, but old prefix was still in the table, although I was not announcing it any more... Q: Do you see any difference between number of announcements vs number of allocations per region? A: This was not researched, it's a good idea for the next time. Bernard Tuy (BT): While discussing the filters - we are at the point when we do need some recommendations from the RIPE community. DK: 6bone/IETF docs exist. Do we need new ones? GD: They need re-doing, can not be accepted by default. DK: Do we need recommendations? GD: We need to at least keep an eye on this, and think about it. We also need to think about multihoming issue. Let's put it on the agenda for the next meeting. DK: Any opinions right now? GD: Just one customer is doing it - "announcing more specific prefix" ?? I find it a good solution. Francis Dupont (FD): That does not scale. It is too easy. ISPs do not want to buy GB of memory to support this. But there's no good alternative solution. GD: Multihoming needs to be done by BGP : either /32, or /48 that can get filtered. There are two groups busy with it in the IETF, official&dead, and another one... Proposal for the presentation on the next meeting - "Approaches to multihoming and why everything sucks". FD: "ipv6mh" is good place to discuss this. It's a closed mailing list. DK: Personally, I think the way Gert is doing it is a good way. DK: Anyone who is a good candidate to lead the discussion, present existing solutions and explain why they don't work, please volunteer for the next meeting. H. Report(s) ============ About *actual* v6 traffic volume as compared to v4? What's *real* out there, not what's on power-point? H.1 baseline measurement for ipv6.int queries ---------------------------------------------- (Terry Manderson, APNIC) [URL - Presentation missing!] Q: How up to date are your graphs, and can you see RIPE43 meeting influence? A: From last Friday, so can't see this meeting. Q: different considerations - DNS traffic as protocol, and v6 traffic after DNS queries. how to get v6 related RRs. unless you have v6 from the root, except the glue, you don't know where else to go. Q: Can you see where are the queries coming from? A: These were the baseline measurements, not analysis. But, most cam from Japan. MS:.jp & .fr have been supporting v6 in DNS for years. We are asking for glue records in root, but according to the policy that is not possible. We are again asking RIPE NCC to support our efforts to tell to the root - v6 is harmless. DK:In the Activity Plan input to the RIPE NCC, we should be making it clear that this is important to us, as members of the RIPE community. [coffee break] H.2 some snap-shot info ---------------------- (Wilfried Woeber) [URL - Presentation missing!] DK: For the next RIPE meeting we will be preparing dns/v6/sec workshop. H.3 Discussion -------------- Do we want some more activity under this topic - to cover it in depth; to have more data? e.g TTM with more data. Since this is a _working_ group, more work from the participants is appreciated... I. Developments/initiatives =========================== regarding IPv6 in the RIPE region and beyond (input from the audience) I.1 Bernard Tuy, Renater ------------------------ We are deploying M6bone. 40 players, 15 non-fr countries. If you are interested, you are more then welcome to join us. http://sem2.renater.fr/m6bone/ What do we want to achieve, what kind of SW is used... I.2 Robert Blechinger, CyberNet ------------------------------- Number of queries per day, on their DNS server: 28 6bone, 2 6to4, 336 RIPE NCC allocated: 211 own, 115 from one peer. I.3 Mohsen Souissi Ready to publish [...] to draft standard status. I.4. DK The same week that there is a RIPE meeting in Barcelona in May there is a IPv6 Forum in Madrid. We are trying to schedule that at least IPv6-wg does not overlap. MS: This might be an opportunity to test video-conferencing over IPv6. Z. AOB ====== Z.1. input to the RIPE NCC Activity Plan ----------------------------------------- Please let your representatives tell to the RIPE NCC management and Executive Board your requirements, so that they can be included in the Activiy Plan. Z.2. 6bone transition to the RIRs ---------------------------------- There has been a talk in the LIR-WG regarding that. 6bone addresses are currently allocated by Bob Fink, which is inconvenient as being a single point of failure, and is outside of the RIR structure. If interested, make sure to take part in the discussion on the lir-wg list & meetings. The progres of the co-ordination efforts will be reported on the next meeting. The most important are the technical issues - ip6.int & ip6.arpa delegations. They will be put on the agenda for the next meeting. Z.3. Other ---------- Info: Bernard Tuy: 29 & 30 October, IPv6 conference in Paris 28 October, IPv6 Tutorial by Mohsen Souissi, like the one done on Monday in Rhodes. Andrew Miehs, CyberNEt: ipv6-wg meeting should not be at the same time EIX!! One of the issues - we'd like to establish peering points, but I can't express this in the IEX-WG. DK: Other people asked for this too, many many times. The same - DNS-WG also has many v6 topics now, and is also scheduled at the same time! WW: Instead of complaining, it's better to put a proposal - how can we improve the scheduling? Action point on DK: to bring the issue to the WG chair lunch, and try to find a solution.

1 0

[ipv6-wg@ripe.net] arpa zone for 2002::/16 ?
by Andrius Kasparavicius 08 Jan '03

08 Jan '03

Here you are talking about 3ffe & arpa. I would like to know if there is something going to make reverse available for 2002::/16 addr space? Maybe there could be used automatic zone delegation by current ipv4 ns delegation? best regards, Andrius

4 4

[ipv6-wg@ripe.net] IPv6 reverse DNS meeting at last IETF
by David Kessens 08 Jan '03

08 Jan '03

Hi, I would like to share with you the results of a private discussion that started at a meeting at the last IETF regarding IPv6 reverse delegation issues. The group of people was an ad-hoc group which is worried that the process for moving to ip6.arpa is stagnating. People involved in the discussion were: Kurt Erik Lindqvist David Kessens Johan Ihren Andrei Robachevsky Patrik Faltstrom Jim Reid The topic of the meeting was IPv6 reverse DNS. We agreed on the following: - the move to ip6.arpa is not on track - it's probably best that we start to give attention to this problem in the DNS wg to avoid any duplication in efforts. (BTW this also means that the chairs of the ipv6 wg and DNS wg need a meeting schedule that will allow IPv6 wg participants to attend the DNS wg - we, Jim Reid & David Kessens, have already made this very clear to the RIPE NCC). - Andrei made it clear that the RIPE NCC wants to move forward with a pragmatic approach - we should stop making further delegations in ip6.int (in so far that this is within our powers, at the minimum we can recommend this) - we should make it possible that 3ffe address holders can start getting reverse delegations in ip6.arpa. This is important because it will allow software coders/vendors to do look ups in ip6.arpa only instead of doing kludges by looking in both or in the wrong one (ip6.int) we think this is important to get started as soon as possible since software tends to stay alive longer than desirable Andrei/RIPE NCC was going to ask IAB and/or other powers that be to get this moving. We as a community should help if necessary by a supporting statement. (Andrei already sent his request and is working on a proposal for implementation) - 6to4 reverse delegation We discussed this topic briefly as a lesser priority item. As 6to4 is a transition mechanism for networks that can not get native access to v6, we concluded that reverse delegation might not be that important for the 6to4 addresses. This is clearly a point for more discussion for the DNS wg. Obviously, this is just the opinion of a small group - your comments are welcome! Please follow up to the dns-wg list only (unless you don't agree on moving this topic to the dns-wg in the first place :-)). Thanks, David K. ---

5 6

[ipv6-wg@ripe.net] Action points from the RIPE 43 meeting
by Andrei Robachevsky 03 Jan '03

03 Jan '03

Dear Colleagues, Please find below a report regarding the status of the action point from the RIPE 43 ipv6-wg meeting. Regards, Andrei Robachevsky RIPE NCC E. IPv6 capable RR DNS servers ============================== RIPE NCC will provide IPv6 access to the ns.ripe.net by the end of November. D. Action point 42.1: ===================== Investigate the CNAME and other solutions for v6-reverse delegation. This action point came up from the request for support a smooth transition from ip6.int to ip6.arpa. To be done in consistent manner this is being coordinated with other RIRs. Also a technical discussion is planned at the IETF-55 in November. We'll let the community know about the outcome of these discussions. F. Action point 42.2: ===================== Give an overview of 6-to-4 reverse delegation issues This issue requires coordination between the RIRs and is being discussed. More information can be available after the IETF-55 meeting in November.

8 23

[ipv6-wg@ripe.net] ipv6 allocation policy question
by Andrius Kasparavicius 12 Dec '02

12 Dec '02

Hello, I am thinking about IPv6 allocation scheme and I actually do not really understand what mean "allocating more tan one /48 for large subscribers only". There I should calculate HD-Ratio? For example RIPE NCC(for internal use i think) has /42? What it mean? Maybe it is just looking to make good hiararchy in future? Or what? Should I, making /48 allocation to average subscriber, reserve next one or two blocks looking to the future's better hiararchy? Tell your opinion or strict policy. best regards, Andrius

1 0

[ipv6-wg@ripe.net] Draft agenda (v1) for IPv6 wg RIPE 44
by David Kessens 11 Dec '02

11 Dec '02

Hi, Below follows the draft agenda (v1) for the ipv6 wg for RIPE44. This is the first version of the agenda so I very much welcome your input on other/more agenda topics. Please let me know if you have anything that you would like to have covered. Also, I am still looking for speakers on various topics, please let me know if you are willing/able to help out. As for: Action point on DK: to bring the issue of wg scheduling to the WG chair lunch, and try to find a solution. The first part (my part :-)) of this action point has been completed. The scheduling problems with the DNS and EIX wg have been brought to the attention of the schedulers and I hope that we will get a better schedule this time. As an experiment, in order to help to achieve this goal of less scheduling conflicts, I offered to shorten the ipv6 wg session to one 2 hour slot to help create more time in the schedule. Thanks, David K. --- Draft agenda (v1) for the IPv6 Working Group Meeting RIPE44 A. Administrative stuff - appointment of scribe - agenda bashing (David Kessens) B. Status of the 6bone (David Kessens) C. IPv6 version of TTM (Henk Uijterwaal) D. Input for the DNS wg: Action point 42.1: Investigate the CNAME and other solutions for v6-reverse delegation (RIPE NCC) IPv6 capable RR DNS servers - what is needed, what are the issues ?!? (RIPE NCC) Action point 42.2: Give an overview of 6-to-4 reverse delegation issues at RIPE 43 (RIPE NCC) F. tentative: IPv6 state of art in the italian educational and research community. (Gabriella Paolini) G. Global IPv6 routing table status (Gert Doering) H. Workable approaches for ipv6 multihoming in the absence of a better solution than the one we use in the ipv4 Internet (I am looking for speakers) I. Report(s) about *actual* v6 traffic volume as compared to v4? what's *real* out there, not what's on powerpoint? (I am looking for speakers) J. Developments/initiatives regarding IPv6 in the RIPE region and beyond (input from the audience) K. input to the RIPE NCC Activity Plan (input from the audience) Z. AOB ---

1 0

[ipv6-wg@ripe.net] Draft minutes (v1) of ipv6 wg session RIPE43
by David Kessens 11 Dec '02

11 Dec '02

Hi, Please see below for the first draft of the minutes of last meeting. Correction/comments are very much welcome. The minutes will be declared final if no major defects (other than typos etc.) are found by Friday, December 13th, 2002. David K. --- Draft minutes (v1) ipv6 wg RIPE 43 Rhodes, Greece 11 September 2002 A. Administrative stuff ======================= Agenda is available at: http://www.kessens.com/~david/presentations/ The scribe was appointed, and the agenda was agreed upon. B. Status of the 6bone (David Kessens (DK)) ====================== Presentation is available at: http://www.kessens.com/~david/presentations/ Per 20020911, there are now 1187 ipv6 sites in 58 different countries on the ipv6 Internet. C. IPv6 version of TTM (Henk Uijterwaal (HU)) ====================== Presentation is available at: http://www.ripe.net/ripe/meetings/archive/ripe-43/ presentations/ripe43-ipv6-ttm/ Wilfried Woeber (WW): With 6net hat - is possible to make preliminary measurements available before the end of the year? DK: How is this service supported from the commercial point of view? Right now, you have to pay for the test-box. HU: You still have to buy a test box, and we operate it for you. Service fee is currently 3000 EUR. Q: Do you get v4 measurements with that, too? A: In principle, yes. WW: Technical question: the ntp protocol is used to encode v4 from clock-host partners to prevent forming loops. Anyone here knows more? HU: forward me the emails from the discussion. DK: I like this new service. This can be a reason for me to order such a box, because v6 production is important for us, and I want to be able to measure the performance. D. Action point 42.1: ===================== Investigate the CNAME and other solutions for v6-reverse delegation E. IPv6 capable RR DNS servers ============================== What is needed, what are the issues ?!? (RIPE NCC) F. Action point 42.2: ===================== Give an overview of 6-to-4 reverse delegation issues at RIPE 43 (RIPE NCC) Andrei Robachevski, RIPE NCC covered topic D, E & F. In regards to D & E, a prototype was made; setups how to support them was tested; no final decision yet. Status: in progress. It will be accomplished before the next ripe meeting Mohsen Souissi (MS): Are you aware of the technical notes, published last week, about DNAME, by the IETF committee? This is only meant for the smooth transition from .int -> .arpa Things are moving very very slowly. There is no dns server for ip6.arpa. "ns.ripe.net" is not ipv6-capable. I have already suggested several times to IANA that AFNIC server is available to be a secondary for ip6.arpa, but I received no answer... DK: There is an agenda point in the AOB about the Activity Plan of the RIPE NCC. We need to add this in there. My suggestion is that the RIPE NCC members make it clear to the RIPE NCC and the RIPE NCC Board that we want RIPE NCC to give some real attention to this. G. Global IPv6 routing table status (Gert Doering (GD)) =================================== Presentation is available at: http://www.ripe.net/ripe/meetings/archive/ripe-43/ presentations/ripe43-ipv6-table/ Q: Do you see people changing over, or announcing both? A: It was possible to see everything, all combinations. Q: Are the filtering problems still there? A: Not really, but old prefix was still in the table, although I was not announcing it any more... Q: Do you see any difference between number of announcements vs number of allocations per region? A: This was not researched, it's a good idea for the next time. Bernard Tuy (BT): While discussing the filters - we are at the point when we do need some recommendations from the RIPE community. DK: 6bone/IETF docs exist. Do we need new ones? GD: They need re-doing, can not be accepted by default. DK: Do we need recommendations? GD: We need to at least keep an eye on this, and think about it. We also need to think about multihoming issue. Let's put it on the agenda for the next meeting. DK: Any opinions right now? GD: Just one customer is doing it - "announcing more specific prefix" ?? I find it a good solution. Francis Dupont (FD): That does not scale. It is too easy. ISPs do not want to buy GB of memory to support this. But there's no good alternative solution. GD: Multihoming needs to be done by BGP : either /32, or /48 that can get filtered. There are two groups busy with it in the IETF, official&dead, and another one... Proposal for the presentation on the next meeting - "Approaches to multihoming and why everything sucks". FD: "ipv6mh" is good place to discuss this. It's a closed mailing list. DK: Personally, I think the way Gert is doing it is a good way. DK: Anyone who is a good candidate to lead the discussion, present existing solutions and explain why they don't work, please volunteer for the next meeting. H. Report(s) ============ About *actual* v6 traffic volume as compared to v4? What's *real* out there, not what's on power-point? H.1 baseline measurement for ipv6.int queries ---------------------------------------------- (Terry Manderson, APNIC) [URL - Presentation missing!] Q: How up to date are your graphs, and can you see RIPE43 meeting influence? A: From last Friday, so can't see this meeting. Q: different considerations - DNS traffic as protocol, and v6 traffic after DNS queries. how to get v6 related RRs. unless you have v6 from the root, except the glue, you don't know where else to go. Q: Can you see where are the queries coming from? A: These were the baseline measurements, not analysis. But, most cam from Japan. MS:.jp & .fr have been supporting v6 in DNS for years. We are asking for glue records in root, but according to the policy that is not possible. We are again asking RIPE NCC to support our efforts to tell to the root - v6 is harmless. DK:In the Activity Plan input to the RIPE NCC, we should be making it clear that this is important to us, as members of the RIPE community. [coffee break] H.2 some snap-shot info ---------------------- (Wilfried Woeber) [URL - Presentation missing!] DK: For the next RIPE meeting we will be preparing dns/v6/sec workshop. H.3 Discussion -------------- Do we want some more activity under this topic - to cover it in depth; to have more data? e.g TTM with more data. Since this is a _working_ group, more work from the participants is appreciated... I. Developments/initiatives =========================== regarding IPv6 in the RIPE region and beyond (input from the audience) I.1 Bernard Tuy, Renater ------------------------ We are deploying M6bone. 40 players, 15 non-fr countries. If you are interested, you are more then welcome to join us. http://sem2.renater.fr/m6bone/ What do we want to achieve, what kind of SW is used... I.2 Robert Blechinger, CyberNet ------------------------------- Number of queries per day, on their DNS server: 28 6bone, 2 6to4, 336 RIPE NCC allocated: 211 own, 115 from one peer. I.3 Mohsen Souissi Ready to publish [...] to draft standard status. I.4. DK The same week that there is a RIPE meeting in Barcelona in May there is a IPv6 Forum in Madrid. We are trying to schedule that at least IPv6-wg does not overlap. MS: This might be an opportunity to test video-conferencing over IPv6. Z. AOB ====== Z.1. input to the RIPE NCC Activity Plan ----------------------------------------- Please let your representatives tell to the RIPE NCC management and Executive Board your requirements, so that they can be included in the Activiy Plan. Z.2. 6bone transition to the RIRs ---------------------------------- There has been a talk in the LIR-WG regarding that. 6bone addresses are currently allocated by Bob Fink, which is inconvenient as being a single point of failure, and is outside of the RIR structure. If interested, make sure to take part in the discussion on the lir-wg list & meetings. The progres of the co-ordination efforts will be reported on the next meeting. The most important are the technical issues - ip6.int & ip6.arpa delegations. They will be put on the agenda for the next meeting. Z.3. Other ---------- Info: Bernard Tuy: 29 & 30 October, IPv6 conference in Paris 28 October, IPv6 Tutorial by Mohsen Souissi, like the one done on Monday in Rhodes. Andrew Miehs, CyberNEt: ipv6-wg meeting should not be at the same time EIX!! One of the issues - we'd like to establish peering points, but I can't express this in the IEX-WG. DK: Other people asked for this too, many many times. The same - DNS-WG also has many v6 topics now, and is also scheduled at the same time! WW: Instead of complaining, it's better to put a proposal - how can we improve the scheduling? Action point on DK: to bring the issue to the WG chair lunch, and try to find a solution.

1 0

[ipv6-wg@ripe.net] RIPE NCC's DNS server available in IPv6
by Daniel Diaz 09 Dec '02

09 Dec '02

Dear all, We are glad to announce that since November 29, RIPE NCC's DNS service (ns.ripe.net) is also available in IPv6. This was an action on RIPE NCC for RIPE 43 and it is now accomplished. ns.ripe.net internet address = 193.0.0.193 ns.ripe.net IPv6 address = 2001:610:240:0:193::193 Best regards. -- Daniel.Diaz Operations Manager RIPE NCC

3 2