have we reached 1000 pre xes yet?gert/ripe/r52-v6-table.pdf · 80 90 05-01-01 05-04-01 05-07-01...

IPv6 routing table Introduction 1'

&

$

%

Have We Reached 1000 Prefixes Yet?

A snapshot of the global IPv6 routing table

Gert Doring, SpaceNet AG, Munich, Germany

April 26st, 2006

RIPE 52, Istanbul

IPv6 routing table Overview 2'

&

$

%

Overview

• numbers

• pictures & trends

• things that should not be there. . .

• conclusions & recommendations

• references

Slides online at: http://www.space.net/ gert/RIPE/R52-v6-table/

IPv6 routing table Numbers 3'

&

$

%

Numbers - AS numbers

• as of 2006/04/24: 589 unique AS numbers visible (06/10: 563)

– 391 origin-only ASes (no transit paths seen) (375)

– 182 ASes originate & give transit (175)

– 16 transit-only ASes (e.g. 2153, 3856, 4774, 6667, . . . ) (13)

• mixture of RIR (2xxx::) and 6Bone (3FFE::) space announced

– 447 ASes originate 1 RIR prefix (408)

– 25 ASes originate 1 6Bone prefix (35)

– 41 ASes originate 1 6Bone + 1 RIR prefix (44)

– 31 ASes originate 2 RIR prefixes (4 due to /32+/35)

– 29 ASes with “more than that”, maximum is 7 prefixes

• 5 ASes still announce their prefix as /32 and /35

• note: all paths observed from AS5539


&

$

%

ASes - why are people announcing 2 prefixes?

• 6bone to RIR migration: 1 6bone, 1 RIR prefix, temporary

2001:420::/35 109 i

3FFE:C00::/24 109 i

• /35 to /32 migration: 2 RIR prefixes, temporary

2001:258::/32 2914 2510 i

2001:258::/35 2914 2510 i

• sub-allocations to (non?-)multihomed customers?2001:388::/32 7575 i

2001:388:1000::/48 7575 18062 i

2001:388:1002::/48 7575 18062 i

2001:388:2::/48 7575 18062 i

2001:388:3002::/48 7575 18062 i

2001:388:6002::/48 7575 18062 i

2001:388:608C::/48 7575 18062 i

2001:388:A000::/40 7575 18062 i

• mergers and acquisitions, business units, growth, . . .2001:360::/32 1221 i

2001:8000::/20 1221 i


&

$

%

Numbers - Prefixes

As of 2006/04/24: 724 prefixes in total (2005/10/11: 720)

/n global RIR space 6bone 6to4 (2005/10/11)/16 1 0 0 1 (1 0 0 1)

/19-21 6 6 0 0 (5 5 0 0)/24 30 0 30 0 (34 0 34 0)/27 1 1 0 0 (1 1 0 0)/28 22 1 21 0 (33 1 32 0)

/29-/30 3 3 0 0 (2 2 0 0)/32 538 513 25 0 (494 468 26 0)

/33-/34 4 4 0 0 (3 3 0 0)/35 24 24 0 0 (24 24 0 0)

/36-/39 1 1 0 0 (3 2 1 0)/40 6 5 1 0 (11 10 1 0)

/41-/47 3 3 0 0 (2 2 0 0)/48 84 77 7 0 (101 88 13 0)

/52-/60 1 1 0 0 (0 0 0 0)/64 0 0 0 0 (5 3 2 0)

/65-/128 0 0 0 0 (1 1 0 0)


&

$

%

Graphics: Total Prefixes - 4.5 years

200

300

400

500

600

700

800

06/0506/0105/0905/0505/0104/0904/0504/0103/0903/0503/0102/0902/0502/0101/09

global


&

$

%

Graphics: RIR vs. 6Bone Prefixes - 4.5 years

0

100

200

300

400

500

600

700

06/0506/0105/0905/0505/0104/0904/0504/0103/0903/0503/0102/0902/0502/0101/09

RIR space6bone space


&

$

%

Graphics: RIR vs. 6Bone Prefixes - 4 months

0

100

200

300

400

500

600

700

06-05-0106-04-0106-03-0106-02-0106-01-01

<-- +19 /48s from AS17832 SIXNGIX KR 2001:2B8:xx::/48

<-- +30 /48s from AS17832 ^ +/- 52 /48s

from AS17832

without AS17832, things would be _really_ boring... RIR space6bone space


&

$

%

Graphics: The Big Drop (02/2006)

620

630

640

650

660

670

680

690

06-01-14 06-01-21 06-01-28 06-02-04 06-02-11 06-02-18 06-02-25 06-03-04 06-03-11 06-03-18

<-- +19 /48s from AS17832 SIXNGIX KR 2001:2B8:xx::/48

<-- +30 /48s from AS17832

^ +/- 52 /48sfrom AS17832

without AS17832, things would be _really_ boring...

<-- 12 /48s from 2001:3C8:: disappear

13 /32s drop from the table --->(some come back over time)

12 /48 + 10 /32 disappear for 3 days -->

(allmost all from Korea) <--- 21 more specifics from different netblocks disappear (11537 cleanup)

RIR space


&

$

%

where did those /32s go?? (2006/02/14⇒15)

• BGP paths for all dropped /32s contain same elements in tail:Network common path tail

2001:4F0::/32 ... 3320 680 1273 6830 6830 6830 6830 6175 139442001:588::/32 ... 3320 680 1273 12702 29052001:598::/32 ... 3320 680 1273 4436 215482001:5C0::/32 ... 3320 680 1273 6830 6830 6830 6830 6175 330182001:850::/32 ... 3320 680 1273 6830 6830 6830 6830 54242001:970::/32 ... 3320 680 20965 11537 17579 1237 17832 9270 2200 26092001:A68::/32 ... 3320 680 1273 12702 160232001:D58::/32 ... 3320 680 20965 11537 9264 17717 94162001:E08::/32 ... 3320 680 1273 6830 6830 6830 6830 6939 4538 239112001:12E0::/32 ... 3320 680 1273 12956 104292001:1360::/32 ... 3320 680 1273 30071 105862001:1598::/32 ... 3320 680 3257 90092001:1B38::/32 ... 3320 680 1273 13193 8554

• looks like a major ghosting problem 680 ⇒ 3302

• BGP-Session 680 → 3320 reset at 2006/02/14 16:00

• IOS involved: 12.3(11)T7 - no “old and unmaintained box”!


&

$

%

Graphics: trends? (12 months)

540

560

580

600

620

640

660

680

700

06-05-0106-03-0106-01-0105-11-0105-09-0105-07-0105-05-01

RIR spacelinear growth?

stagnation?linear again?The Big Drop

apparent stagnation?

-> caused by disappearanceof some APNIC prefixes


&

$

%

Numbers: RIRs, Allocations, . . .

• On 2006/04/24, 1056 LIR blocks (2000::/4) allocated by RIRs:

RIR alloc. members perc. on 2005/10/09

ARIN 198 ˜ 2440 8.1% 169 (+17%)

APNIC 249 ˜ ?1890 13.2% 219 (+13%)

RIPE 547 ˜ 4311 12.7% 497 (+10%)

LACNIC 51 ˜ 561 9.1% 39 (+30%)

AfriNIC 11 ˜ 810 1.4% -

• note: not counting /48 microallocs and /35⇒/32 extentions

• actual percentage with IPv6 similar for RIPE and APNIC

• 437 (R51: 483) allocations visible in routing table (only 41%!)


&

$

%

Numbers: RIRs: notable allocations (1)

• more “very large” allocations seen:– 2001:44d0::/28 Korea Cable Television, KR (2005/11/10)

– 2001:4510::/29 Beijing ShenZhou Greatwall, CN (2006/01/23)

– 2001:4540::/27 Taiwan Fixed Network Co., TW (2005/12/22)

– 2001:4580::/26 Seednet Digital United, TW (2005/12/12)

– 2001:b000::/21 HiNet Taiwan, TW (2006/03/15)

– 2404::/26 Asia Pacific Online Service, TW (2006/01/25)

– 2404:80::/28 SONET Taiwan, TW (2006/01/25)

– 2800:20::/28 Comsat Argentina, AR (2005/12/14)

– 2a01:b0::/31 IT Solid Solutions, BE (2006/04/13)

– 2a01:800::/23 Vodafone Deutschland, DE (2006/04/12)

– 2a01:1000::/21 Polish Telecom, PL (2006/02/01)

– 2a01:c000::/19 France Telecom, FR (2005/12/30)

– 2a01:00b8::/32 Vatican City State, VA (2006/04/18) *

• ⇒ check your BGP filters!!


&

$

%

Numbers: RIRs: notable allocations (2)

• Allocations ICANN ⇒ RIRs since RIPE 51

Prefix RIR Date Comment

2001:4400::/23 AfriNIC (reassigned)

2001:B000::/20 APNIC 08 Mar 06 TW/HiNet /21

2404:0000::/23 APNIC 19 Jan 06

2610:0000::/23 ARIN 17 Nov 05

2800:0000::/23 LACNIC 17 Nov 05

2A01:0000::/16 RIPE NCC 15 Dec 05 PL/21, FR/19

• http://www.iana.org/assignments/ipv6-unicast-address-assignments


&

$

%

Graphics: prefixes by RIR region

0

50

100

150

200

250

300

350

400

06-04-0106-01-0105-10-0105-07-0105-04-0105-01-01

The Big Drop -+|v

RIPEAPNIC

ARINLACNICAfriNIC6bone


&

$

%

Graphics: prefixes by country (RIPE)

0

10

20

30

40

50

60

70

80

90

06-04-0106-01-0105-10-0105-07-0105-04-0105-01-01

EUDEGBNLFRIT

CH


&

$

%

Graphics: prefixes by country (APNIC)

0

10

20

30

40

50

60

70

80

90

06-04-0106-01-0105-10-0105-07-0105-04-0105-01-01

The Big Drop -+(Korean part) |

|v

JPKRTWCNAUMYIN


&

$

%

Graphics: Allocated vs. Routed

0

50

100

150

200

250

300

1999 2000 2001 2002 2003 2004 2005 2006

allocated/yr.routed 2002/01routed 2003/01routed 2004/01routed 2005/01routed 2006/01

routed 06/04/24


&

$

%

Why are those prefixes not visible?

• looking at prefixes allocated in 2004 (255)

• visible today: 121, at some point: 25, never: 109

• categorizing 109 prefixes never seen in routing tables:

– by region: APNIC: 24, ARIN: 26, LACNIC: 2, RIPE: 57

– by size: 1x /21, /24, /27, 106x /32

– (specifically not looking at IXP /48s)

– by organizational type (guessing from whois)

∗ commercial 105

∗ research 2 (.in, .kr)

∗ military/governmental 2 (CNY, US DoT)

IPv6 routing table Weirdos 20'

&

$

%

Graphics: route6 objects vs. routes seen

0

50

100

150

200

250

300

350

400

04/09 04/11 05/01 05/03 05/05 05/07 05/09 05/11 06/01 06/03 06/05

2004/12/29: RPSLng support

in RIPE DB --+|v

RIPE routesRIPE route6 obj

IPv6 routing table route6 is good for you 21'

&

$

%

route6 object example

• it’s as easy as this...

route6: 2001:608::/32

descr: DE-SPACE-2001-0608

descr: SpaceNET AG, Munich

origin: AS5539

notify: [email protected]

mnt-by: SPACENET-N

changed: [email protected] 20041230

source: RIPE

• strongly recommended, helps upstream/peer ASes build decent

BGP filters, based on IRR data

IPv6 routing table News 22'

&

$

%

miscellaneous news

• 2006/06/06 is the offical end of the 6bone (next month!)

– ⇒ 3FFE address allocations are no longer valid anymore

– consequences for the routing table? filtering police?

• 2001::/32 assigned for Teredo routing on January 10, 2006

– expect “inconsistant” announcements, as for 2002::/16

route6: 2001:0000::/32

descr: Teredo-ITGate

remarks: Teredo anycast route. See TEREDO-MNT for details.

origin: AS12779

mnt-by: ITGATE-NCC

mnt-routes: TEREDO-MNT

IPv6 routing table Conclusions 23'

&

$

%

References

• Ghost Route Hunter: http://www.sixxs.net/tools/grh/

• List of IPv6 blocks allocated by the RIRs:

http://www.ripe.net/rs/ipv6/stats/index.html

• MIPP (minimum peering policy) project:

http://ip6.de.easynet.net/ipv6-minimum-peering.txt

• IPv6 sample prefix filter page

http://www.space.net/ gert/RIPE/ipv6-filters.html

• Slides are available at:

http://www.space.net/ gert/RIPE/R52-v6-table/

IPv6 routing table Conclusions 24'

&

$

%

Questions?

[email protected]

have we reached 1000 pre xes yet?gert/ripe/r52-v6-table.pdf · 80 90 05-01-01 05-04-01 05-07-01...

Documents