idnits 2.17.1 draft-song-yeti-testbed-experience-00.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack a Security Considerations section. ** The document seems to lack separate sections for Informative/Normative References. All references will be assumed normative when checking for downward references. ** There are 19 instances of too long lines in the document, the longest one being 19 characters in excess of 72. == There are 21 instances of lines with non-RFC2606-compliant FQDNs in the document. == There are 2 instances of lines with non-RFC6890-compliant IPv4 addresses in the document. If these are example addresses, they should be changed. == There are 2 instances of lines with non-RFC3849-compliant IPv6 addresses in the document. If these are example addresses, they should be changed. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (November 20, 2015) is 3073 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Unused Reference: 'RFC1035' is defined on line 802, but no explicit reference was found in the text ** Obsolete normative reference: RFC 2460 (Obsoleted by RFC 8200) ** Obsolete normative reference: RFC 7626 (Obsoleted by RFC 9076) Summary: 5 errors (**), 0 flaws (~~), 5 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Internet Engineering Task Force L. Song 3 Internet-Draft S. Kerr 4 Intended status: Informational D. Liu 5 Expires: May 23, 2016 Beijing Internet Institute 6 November 20, 2015 8 Experiences from Root Testbed in the Yeti DNS Project 9 draft-song-yeti-testbed-experience-00 11 Abstract 13 This document reports and discusses issues in DNS root services, 14 based on experiences from the experiments in the Yeti DNS project. 15 These issues include IPv6-only operation, the root DNS server naming 16 scheme, DNSSEC KSK rollover, root server renumbering, multiple root 17 zone signer, and so on. This project was founded in May 2015 and has 18 since built a live root DNS server system testbed with volunteer root 19 server and resolver operations. 21 Status of This Memo 23 This Internet-Draft is submitted in full conformance with the 24 provisions of BCP 78 and BCP 79. 26 Internet-Drafts are working documents of the Internet Engineering 27 Task Force (IETF). Note that other groups may also distribute 28 working documents as Internet-Drafts. The list of current Internet- 29 Drafts is at http://datatracker.ietf.org/drafts/current/. 31 Internet-Drafts are draft documents valid for a maximum of six months 32 and may be updated, replaced, or obsoleted by other documents at any 33 time. It is inappropriate to use Internet-Drafts as reference 34 material or to cite them other than as "work in progress." 36 This Internet-Draft will expire on May 23, 2016. 38 Copyright Notice 40 Copyright (c) 2015 IETF Trust and the persons identified as the 41 document authors. All rights reserved. 43 This document is subject to BCP 78 and the IETF Trust's Legal 44 Provisions Relating to IETF Documents 45 (http://trustee.ietf.org/license-info) in effect on the date of 46 publication of this document. Please review these documents 47 carefully, as they describe your rights and restrictions with respect 48 to this document. Code Components extracted from this document must 49 include Simplified BSD License text as described in Section 4.e of 50 the Trust Legal Provisions and are provided without warranty as 51 described in the Simplified BSD License. 53 Table of Contents 55 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 56 2. Problem Statement . . . . . . . . . . . . . . . . . . . . . . 3 57 3. Yeti Testbed and Experiment Setup . . . . . . . . . . . . . . 4 58 3.1. Distribution Master . . . . . . . . . . . . . . . . . . . 6 59 3.1.1. Yeti root zone SOA SERIAL . . . . . . . . . . . . . . 6 60 3.1.2. Timing of Root Zone Fetch . . . . . . . . . . . . . . 7 61 3.1.3. Information Synchronization . . . . . . . . . . . . . 7 62 3.2. Yeti Root Servers . . . . . . . . . . . . . . . . . . . . 8 63 3.3. Yeti Resolvers and Experimental Traffic . . . . . . . . . 10 64 4. Experiments in Yeti Testbedd . . . . . . . . . . . . . . . . 10 65 4.1. Naming Scheme and Glue Issue . . . . . . . . . . . . . . 11 66 4.2. Multiple-Signers with Multi-ZSK . . . . . . . . . . . . . 12 67 4.3. Root Renumbering Issue and Hint File Update . . . . . . . 14 68 4.4. DNS Fragments . . . . . . . . . . . . . . . . . . . . . . 15 69 4.5. The KSK Rollover Experiment in Yeti . . . . . . . . . . . 15 70 5. Other Technical findings and bugs . . . . . . . . . . . . . . 16 71 5.1. IPv6 fragments issue . . . . . . . . . . . . . . . . . . 16 72 5.2. Root compression issue . . . . . . . . . . . . . . . . . 17 73 6. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 17 74 7. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 17 75 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 17 76 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 19 78 1. Introduction 80 RFC 1034[RFC1034] says the domain name space is a tree structure. 81 The top level of the tree for the unique identifier system is the DNS 82 root system. It has been operational for 25+ years. It is pivotal 83 to making the current Internet useful. So it is considered somewhat 84 ossified for stability reasons. It is hard to test and implement new 85 ideas evolving to a more advanced level to counter challenges like 86 IPv6-only operation, DNSSEC key/algorithm rollover [RFC4986], scaling 87 issues, and so on. In order to make the test more practical, it is 88 also necessary to involve users' environments which are highly 89 diversified, in order to study the effect of the changes in question. 91 To benefit Internet development as a whole, the Yeti Project 92 [Yeti-DNS-Project] was proposed to build a parallel, experimental, 93 live IPv6 DNS root system to discover the limits of DNS root name 94 service and deliver useful technical output. Possible research 95 agenda will be explored on this testbed, covering several aspects 96 (but not limited to): 98 o IPv6-only operation 100 o DNSSEC key rollover 102 o Renumbering issues 104 o Scalability issues 106 o Multiple zone file signers 108 Starting from May 2015, three coordinators began to build this live 109 experimental environment and called for participants. At the time of 110 writing, there are 14 Yeti root servers with 13 operators, and 111 experimental traffic from volunteers, universities, DNS vendors, 112 mirrored traffic non-Yeti traffic, and RIPE Atlas probes. Some 113 experiments have been proposed and have been verified in lab tests. 115 Note that the Yeti DNS project has complete fealty to IANA as the DNS 116 name space manager. All IANA top-level domain names will be 117 precisely expressed in the Yeti DNS system, including all TLD data 118 and meta-data[Root-Zone-Database]. So, the Yeti DNS project is not 119 an "alternative root" in the usual sense of that term. It is 120 expected to inform the IANA community by peer-reviewed science as to 121 future possibilities to consider for the IANA root DNS system. 123 In order to let people know the technical activities in Yeti DNS 124 project, this document reports and discusses issues on root DNS 125 services, based on experiences so far from the experiments in the 126 Yeti DNS project. 128 2. Problem Statement 130 Some problems and policy concerns over the DNS Root Server system 131 stem from centralization from the point of view of DNS content 132 consumers. These include external dependencies and surveillance 133 threats. 135 o External Dependency. Currently, there are 12 DNS Root Server 136 operators for the 13 Root Server letters, with more than 500 137 instances deployed globally. Yet compared to the number of 138 connected devices, AS networks, and recursive DNS servers, the 139 number of root instances is far from sufficient. Connectivity 140 loss between one autonomous network and the IANA root name servers 141 usually results in loss of local service within the local network, 142 even when internal connectivity is perfect 144 o Surveillance risk. Even when one or more root name server anycast 145 instances are deployed locally or in a nearby network, the queries 146 sent to the root servers carry DNS lookup information which 147 enables root operators or other parties to analyze the DNS query 148 traffic. This is a kind of information leakage [RFC7626] which is 149 to some extent not acceptable to some policy makers 151 People are often told that the current root system with 13 root 152 servers is not able to be extended to alleviate the above concerns, 153 because it is limited to 13 by the current DNS protocol [ROOT-FAQ]. 154 To the best of author's knowledge, there is no scientific evidence to 155 support this assertion. It remains an open question. 157 There are some technical issues in the areas of IPv6 and DNSSEC, 158 which were introduced to the DNS root server system after it was 159 created. Renumbering DNS root servers also creates some technical 160 issues. 162 o IPv6-only capability. Currently some DNS servers including root 163 which support both A and AAAA (IPv4 and IPv6) records still do not 164 respond to IPv6 queries. IPv6 introduces larger IP packet MTU 165 (1280 bytes) and a different fragmentation model [RFC2460]. It is 166 not clear whether DNS can survive without IPv4 (in an IPv6-only 167 environment), or what the impact of IPv6-only environment 168 introduces to current DNS operations especially in the DNS root 169 server system. 171 o KSK rollover. Currently, IANA rolls the ZSK every six weeks but 172 the KSK has never been rolled as of writing. Is the 512 bytes DNS 173 packet size limitation still observed? Is [RFC5011] widely 174 supported by resolvers? How about longer key with different 175 encryption algorithm? There are many issues still unknown. 177 o Renumbering issue. It is likely that root operators may change 178 their IP addresses for root servers as well. There is no dynamic 179 update mechanism to inform resolvers and other Internet 180 infrastructure relying on root service of such changes. 182 3. Yeti Testbed and Experiment Setup 184 To use the Yeti testbed operationally, the information that is 185 required for correct root name service is a matching set of the 186 following: 188 o a root "hints file" 190 o the root zone apex NS record set 192 o the root zone's signing key 193 o root zone trust anchor 195 Although Yeti DNS project publishes strictly IANA information for TLD 196 data and meta-data, it is necessary to use a special hint file and 197 replace the apex NS RRset with Yeti authority name servers, which 198 will enable the resolves to find and stick to the Yeti root system. 199 In addition, unless IANA was to help Yeti sign its root zone with a 200 different root set, it is necessary to use a different ZSK and KSK 201 (the DNSSEC trust anchor) in Yeti system. 203 Below is a figure to demonstrate the topology of Yeti and the basic 204 data flow, which consists of the Yeti distribution master, Yeti root 205 server, and Yeti resolver: 207 +------------------------+ 208 | IANA Root Zone via | 209 +-+ F.root-servers.net +--+ 210 | +-----------+------------+ | 211 +-----------+ | | | IANA Root.Zone 212 | Yeti | | | | 213 | Traffic | +--v---+ +---v--+ +-----v+ 214 | Collection| | BII | | WIDE | | TISF | 215 | | | DM | | DM | | DM | 216 +---+----+--+ +------+ +-+----+ +---+--+ 217 ^ ^ | | | 218 | | | | | Yeti Root.Zone 219 | v v v 220 | 221 | +------+ +------+ +------+ 222 +- -+ Yeti | | Yeti | ..... | Yeti | 223 | | Root | | Root | | Root | 224 +---+--+ +---+--+ +--+---+ 225 | | | | 226 pcap ^ ^ ^ TLD lookup 227 | upload | | | 229 | +--------------------------+ 230 +- - - - - - - - - -+ Yeti Resolvers | 231 | (with Yeti Hint) | 232 +--------------------------+ 234 Figure 1. The topology of Yeti testbed 236 3.1. Distribution Master 238 As shown in figure 1, the Yeti Root system takes the IANA root zone 239 and performs minimal changes needed to serve the zone from the Yeti 240 root servers instead of the IANA root servers. In Yeti, this 241 modified root zone is generated by the Yeti Distribution Masters 242 (DM), which provide it to the Yeti root servers. 244 So the generation process is: 246 o DM downloads the latest IANA root zone at a certain time 248 o DM makes modifications to change from the IANA to Yeti root 249 servers 251 o DM signs the new Yeti root zone 253 o DM publishes the new Yeti root zone to Yeti root servers 255 While in principle this could be done by a single DM, Yeti uses a set 256 of three DMs to avoid any sense that the Yeti project is run by a 257 single organization. The three Distribution Masters (DMs) can 258 independently fetch the root zone from IANA, sign it and publish the 259 latest zone data to Yeti root servers. 261 In the same while, these DMs coordinate their work so that the 262 resulting Yeti root zone is always consistent. There are two aspects 263 of coordination between three DMs: timing and information 264 synchronization. 266 3.1.1. Yeti root zone SOA SERIAL 268 Consistency with IANA root zone except the top level apex record is 269 one of most important point for the project. As part of Yeti DM 270 design, the Yeti SOA SERIAL which reflect the changes of yeti root 271 zone is one factor to be considered. 273 Currently IANA SOA SERIAL number for root zone is in the form of 274 YYYYMMDDNN, like 2015111801. In Yeti root system, IANA SOA SERIAL is 275 directly copied in to Yeti SOA SERIAL. So once the IANA root zone 276 has changed with a new SOA SERIAL, a new version of the Yeti root 277 zone is generated with the same SOA SERIAL. 279 There is a case of Yeti DM operation that when a new Yeti root server 280 added, DM operator change the Yeti root zone without change the SOA 281 SERIAL which introduces inconsistency of Yeti root system. To avoid 282 inconsistency, the DMs hold on every changes to Yeti apex record and 283 only new IANA SOA SERIAL will trigger the operation of adding these 284 changes to Yeti root zone. 286 A analysis of IANA convention shows IANA SOA SERIAL change 2 times 287 every day (NN=00, 01). And that since October 2007 the maximum of NN 288 was 03 while 13 times it is observed that the versions with NN=02. 289 So in the worst case, the changes of Yeti apex record is updated into 290 Yeti root zone in less than 12 hours. 292 3.1.2. Timing of Root Zone Fetch 294 Yeti root system operators do not receive notify message from IANA 295 when IANA root zone updates with a new SOA serial number. So Yeti 296 DMs check the root zone periodically. At the time of writing, each 297 Yeti DM checks to see if the IANA root zone has changed hourly, on 298 the following schedule: 300 +-------------+---------+ 301 | DM Operator | Time | 302 +-------------+---------+ 303 | BII | hour+00 | 304 | WIDE | hour+20 | 305 | TISF | hour+40 | 306 +-------------+---------+ 308 Note that Yeti DMs can check IANA root zone more frequently (every 309 minute for example). A test done by Yeti participant shows that the 310 delay of IANA root zone update from the first IANA root server to 311 last one is around 20 minute. Once a Yeti DM fetch the new root 312 zone, it will notify all the Yeti root server with a new SOA serial 313 number. So normally yeti root server will be notified in less than 314 20 minute after new IANA root zone generated. Ideally, if IANA DM 315 notifies the Yeti DMs, Yeti root zone will be updated more timely. 317 3.1.3. Information Synchronization 319 Given three DMs operational in Yeti root system, it is necessary to 320 prevent any inconsistency caused by human mistakes in operation. The 321 straight method is to share the same parameters to produce the Yeti 322 root zone. There parameters includes following set of files: 324 o the list of Yeti root servers, including: 326 * public IPv6 address and host name 328 * IPv6 addresses originating zone transfer 330 * IPv6 addresses to send DNS notify to 332 o the ZSKs used to sign the root 334 o the KSK used to sign the root 336 o the SERIAL when this information is active 338 The theory of operation is straight that each DM operator runs a Git 339 repository, containing files with the information needed to produce 340 the Yeti root zone. When a change is desired (such as adding a new 341 server or rolling the ZSK), a DM operator updates the local Git 342 repository. A SOA SERIAL in the future is chosen for when the 343 changes become active. The DM operator then pushes the changes to 344 the Git repositories of the other two DM operators. When the SOA 345 SERIAL of the root zone passes the number chosen, then the new 346 version of the information is used. 348 3.2. Yeti Root Servers 350 In Yeti Root system, authoritative servers donated and operated by 351 Yeti volunteers are configured as a slave to the Yeti DM. As the 352 time of writing, there are 14 Yeti root servers distributed around 353 the world. As one of operational research goal, all authoritative 354 servers are required to work in an IPv6-only environment. In 355 addition, different from IANA root, Yeti root server only serve the 356 Yeti root zone, other than root-servers.org zone and .arpa zone. 358 . 3600000 IN NS bii.dns-lab.net 359 bii.dns-lab.net 3600000 IN AAAA 240c:f:1:22::6 360 . 3600000 IN NS yeti-ns.tisf.net 361 yeti-ns.tisf.net 3600000 IN AAAA 2001:559:8000::6 362 . 3600000 IN NS yeti-ns.wide.ad.jp 363 yeti-ns.wide.ad.jp 3600000 IN AAAA 2001:200:1d9::35 364 . 3600000 IN NS yeti-ns.as59715.net 365 yeti-ns.as59715.net 3600000 IN AAAA 2a02:cdc5:9715:0:185:5:203:53 366 . 3600000 IN NS dahu1.yeti.eu.org 367 dahu1.yeti.eu.org 3600000 IN AAAA 2001:4b98:dc2:45:216:3eff:fe4b:8c5b 368 . 3600000 IN NS ns-yeti.bondis.org 369 ns-yeti.bondis.org 3600000 IN AAAA 2a02:2810:0:405::250 370 . 3600000 IN NS yeti-ns.ix.ru 371 yeti-ns.ix.ru 3600000 IN AAAA 2001:6d0:6d06::53 372 . 3600000 IN NS yeti.bofh.priv.at 373 yeti.bofh.priv.at 3600000 IN AAAA 2a01:4f8:161:6106:1::10 374 . 3600000 IN NS yeti.ipv6.ernet.in 375 yeti.ipv6.ernet.in 3600000 IN AAAA 2001:e30:1c1e:1::333 376 . 3600000 IN NS yeti-dns01.dnsworkshop.org 377 yeti-dns01.dnsworkshop.org 3600000 IN AAAA 2001:1608:10:167:32e::53 378 . 3600000 IN NS yeti-ns.conit.co 379 yeti-ns.conit.co 3600000 IN AAAA 2607:ff28:2:10::47:a010 380 . 3600000 IN NS dahu2.yeti.eu.org 381 dahu2.yeti.eu.org 3600000 IN AAAA 2001:67c:217c:6::2 382 . 3600000 IN NS yeti.aquaray.com 383 yeti.aquaray.com 3600000 IN AAAA 2a02:ec0:200::1 384 . 3600000 IN NS yeti-ns.switch.ch 385 yeti-ns.switch.ch 3600000 IN AAAA 2001:620:0:ff::29 387 Figure 2. the Yeti root server in hint file 389 Since Yeti is a a live root DNS server system testbed, it needs to 390 capture DNS traffic sent for later analysis. Today some servers use 391 dnscap, which is a DNS-specific tool to produce pcap files. There 392 are several versions of dnscap floating around; some people use the 393 VeriSign one. Since dnscap loses packets in some cases (tested on a 394 Linux kernel), some people use pcapdump. It requires the patch 395 attached to this bug report [dnscap-bug-report] 397 System diversity is also a requirement and observed for current 14 398 Yeti root server. Here are the results of a survey regarding the 399 machine, operation system and DNS software: 401 o Machine: 11 out of 14 root server operator are using a VPS to 402 provide service. 404 o OS: 6 operators use Linux (including Ubuntu, Debian, CentOS). 5 405 operators use FreeBSD and 1 NetBSD. 2 other servers are unknown. 407 o DNS software: 8 our of 14 root server use BIND (varying from 9.9.7 408 to 9.10.3). 4 of them use NSD (4.10 and 4.15). The other 2 409 servers use Knot (2.0.1 and 2.1.0). 411 3.3. Yeti Resolvers and Experimental Traffic 413 In client side of Yeti project, we expect participants and volunteers 414 from individual researchers, labs of universities, companies and 415 institutes, and vendors (for example, the DNS software implementers), 416 developers of CPE devices &IoT devices, and middle box developers who 417 can test their product and connect their own testbed into Yeti 418 testbed. Resolvers donated by Yeti volunteers are required to be 419 configured with Yeti hint file and Yeti DNSSEC KSK. It is required 420 that Yeti resolver can speak both IPv4 and IPv6, given that not all 421 the stub resolver and authoritative servers on the Internet are IPv6 422 capable. 424 At the time of writing several universities and labs have joined us 425 and contributed certain amount of traffic to Yeti testbed. But it is 426 far from the desired volume of the experiment traffic. So Yeti 427 adopts two alternative ways to increase the experimental traffic in 428 Yeti testbed to check the functionality of Yeti root system. 430 One approach is to mirror the real traffic by off-path method and 431 reply it into Yeti testbed; this is implemented by one of the Yeti 432 root server operators. Another approach is to use some traffic 433 generating tool such as RIPE Atlas probes to generate specific 434 queries against Yeti servers. 436 4. Experiments in Yeti Testbedd 438 The main goal of Yeti DNS Project is to act as an experimental 439 network. Experiments will be conducted on this network. In order to 440 make the findings that result from these experiments more rigorous, 441 an experiment protocol is proposed. 443 A Yeti experiment goes through four phases: 445 o Proposal. The first step is to make a proposal. It is discussed 446 and if accepted by the Yeti participants then it can proceed to 447 the next phase. 449 o Lab Test. The next phase is to run a version of the experiment in 450 a controlled environment. The goal is to check for problems such 451 as software crashes or protocol errors that may cause failures on 452 the Yeti network, before putting onto the experimental network. 454 o Yeti Test. The next phase actually running the experiment on the 455 Yeti network. Details of this will depend on the experiment. It 456 must be coordinated with the Yeti participants. 458 o Report of Findings. When completed, a report of the findings of 459 the experiment should be made. It need not be an extensive 460 document. 462 In this section, we are going to introduce some experiments 463 implemented and planned in Yeti project. 465 4.1. Naming Scheme and Glue Issue 467 In root server history, the naming scheme for individual root servers 468 was not fixed. Current IANA Root server adopt [a-m].root-servers.net 469 to represent 13 servers which are labeled with letter from A to M. 470 For example, L root operated by ICANN uses l.root-servers.net to 471 represent their server as NS. One reason behind this naming scheme 472 is that the common suffix 'root-servers.net' can be compressed in DNS 473 message to produce a smaller DNS response. 475 Different from the IANA root naming scheme, the Yeti root system uses 476 separate and normal domains for root servers (shown in figure 2). 477 The motivation of this naming scheme in Yeti is that it intentionally 478 produces larger packets for priming responses. Note that currently, 479 the Yeti root has a priming response which is almost the same size as 480 the IANA root. Yeti has no compression, and has one more name 481 server, but it also has no IPv4 addresses. 483 the change of name scheme not only affects the size of priming 484 response, but also changes the content in additional section of the 485 response.When a resolver bootstraps, it sends a 'NS-for-dot' query to 486 one of the root servers that it knows about, which is called a 487 priming query. It looks like this with the "dig" command: 489 $ dig @a.root-servers.net -t ns +norecurse +edns +nodnssec 491 Normally in IANA root system the priming response contains the 492 _names_ of the root servers in the answer section and the _addresses_ 493 of the root servers in the additional section. The additional 494 section data is what the resolvers need to actually perform 495 recursion. Shown as below: 497 In priming response : 499 Answer Section: 500 --------------- 501 . 518400 IN NS a.root-servers.net. 502 . 518400 IN NS b.root-servers.net. 503 ... 504 . 518400 IN NS m.root-servers.net. 506 Addtional section: 507 ------------------ 508 a.root-servers.net. 3600000 IN A 198.41.0.4 509 b.root-servers.net. 3600000 IN A 192.228.79.201 510 ... 511 m.root-servers.net. 3600000 IN AAAA 2001:dc3::35 513 In IANA root system, all the root server returns the "a.root- 514 servers.net" addresses in the additional section, because root 515 servers not only answer for root zone, but also answer for "root- 516 servers.net" zone. Note that BIND will not behave like this if it is 517 not configured for the "root-servers.net" zone. NSD and Knot happily 518 return such "glue" in the additional section, whether configured for 519 the "root-servers.net" zone or not. 521 The Yeti root naming scheme uses separate and independent domain for 522 individual root servers. It this case, the priming response from 523 Yeti root servers will only contain the A&AAAA records of that 524 responding server in BIND 9. However it is desired that the Yeti 525 root servers to respond to priming queries with the addresses of all 526 Yeti root servers in the additional section. This will make them 527 operate as similar to the IANA root servers as possible. 529 In Yeti root system, there are two approaches adopted in different 530 root servers. One is to patch BIND 9 so that it includes the glue 531 addresses in the additional section. The other one is to add a zone 532 file for each root server and answer for all of them at each Yeti 533 server. That means each Yeti root server would have a small zone 534 file for "bii.dns-lab.net", "yeti-ns.wide.ad.jp", "yeti-ns.tisf.net", 535 and so on. 537 4.2. Multiple-Signers with Multi-ZSK 539 According to the Problem statement of Yeti DNS project, more 540 independent participants and operators of root system is desirable. 541 As the name implies, multi-ZSK mode will introduce different ZSKs 542 sharing a single unique KSK, as opposed to the IANA root system 543 (which uses a single ZSK to sign the root zone). On the condition of 544 good availability and consistency on root system, the Multi-ZSK 545 proposal is designed to give each DM operator enough room to manage 546 their own ZSK, by choosing different ZSK, length, duration, and so 547 on; even the encryption algorithm may vary. 549 According to the Yeti experiment protocol, a lab test was done to 550 verify the concept and validity of Multi-ZSK. The purpose of the 551 test is two-fold: 1) To test whether this proposal can be implemented 552 by current DNS protocol&software (to see if there should be some 553 extra modification to protocol or software), and 2) To demonstrate 554 the impact of Multi-ZSK proposal to the current root system. 556 The experiment is run like this: build a test topology like figure 3, 557 with 2 Root servers and a resolver (BIND 9). The hint file of this 558 test only contains the two DM servers. In the first time slot, Root 559 A is up and Root B is turned off. Let resolver bootstrap from Root A 560 and query a certain signed TLD (or junk query). For the second time 561 slot, turn off Root A and turn on Root B. Let resolver shift to Root 562 B to look up another TLD (or a junk query). the test result of 563 different time slot is compared to see whether the resolver can 564 validate the DNSSEC signature. 566 +---------------+ +---------------+ 567 | Root A | | Root B | 568 | (ZSK A) | | (ZSK B) | 569 +-------+-------+ +--------+------+ 570 | | 571 --------+------------+--------------+-------- 572 | 573 +---------+----------+ 574 | | 575 | Resolver | 576 +--------------------+ 578 Figure 3. Multi-ZSK lab test topology 580 There are two cases in this test: 582 o Case 1: Assign one ZSK to the smart sign process on each DM, which 583 means the root zone only contain one single ZSK. 585 o Case 2: Assign both ZSK A and ZSK B to the smart sign process on 586 each DM, which means the root zone contains two ZSK. In this 587 case, it is required that Root A and Root B to share their public 588 ZSK to each other before root zone is signed. 590 In case 1 SERVFAIL is received during switching because the resolver 591 can not validate the signature signed by Root B after switching. In 592 case 2 NOERROR is received. It is the actual demonstration of how 593 Multi-ZSK works by adding multiple ZSK to the root zone. As a 594 result, the resolver will cache the key sets instead of single ZSK to 595 validate the data no matter it is signed by Root A or Root B. As 596 follow-up test, Unbound also passed the test with more than 10 DMs 597 and 10 ZSKs. 599 Although more DM and ZSK can be added into the test, adding more ZSKs 600 to root zone enlarges the DNS response size for DNSKEY queries which 601 may be a concern given the limitation of DNS packet size. Current 602 IANA root server operators are inclined to keep the packets size as 603 small as possible. So the number of DM and ZSK will be parameter 604 which is decided based on operation experience. In the current Yeti 605 root testbed, there will be 3 DMs, each with a separate ZSK. 607 4.3. Root Renumbering Issue and Hint File Update 609 With the nearing renumbering of H root Server's IP address, there is 610 a discussion of ways that resolvers can update their hint file. 611 Traditional ways include using FTP protocol by doing a wget and using 612 dig to get the servers' addresses manually. Each way would depend on 613 operators manual operation. As a result, there are many old machines 614 that have not updated their hint files. As a proof, after done 615 renumbering for thirteen years, there is an observation that the "Old 616 J-Root" can still receive DNS query traffic [Renumbering-J-Root]. 618 This experiment proposal aims to find an automatic way for hint-file 619 updating. The already-completed work is a shell script tool which 620 provides the function that update a hint-file in file system 621 automatically with DNSSEC and trust anchor validation. 623 The methodology is straightforward. The tool first queries the NS 624 list for "." domain and queries A and AAAA record for every domain on 625 the NS list. It requires DNSSEC validation for both signature and 626 trust anchor for all the answers. After getting all the answers, the 627 tool compares the new hint file to the old one. If there is a 628 difference, it renames the old one with a time-stamp and replaces the 629 old one with the new one. Otherwise the tool deletes the new hint 630 file and nothing will be changed. 632 Note that in current IANA root system the servers named in the root 633 NS record are not signed due to lack of incentive. So the tool can 634 not fully work in the production network. In Yeti root system some 635 of the names listed in the NS record are signed, which provides a 636 test environment for such a proposal. 638 4.4. DNS Fragments 640 In consideration of new DNS protocol and operation, there is always a 641 hard limit on the DNS packet size. Take Yeti for example: adding 642 more root servers, using the Yeti naming scheme, rolling the KSK and 643 Multi-ZSK increase the packet size. The fear of large DNS packets 644 mainly stem from two aspects: one is IP-fragments and the other is 645 frequently falling back to TCP. 647 In Yeti testbed, a mechanism is implemented which supports larger DNS 648 packet working around the IP-layer fragment caused by middle box 649 misbehavior (in IPv4) and IPv6 MTU limitation by splitting a single 650 DNS message across multiple UDP datagrams. This DNS fragments 651 mechanism is documented in [I-D.muks-dns-message-fragments] as an 652 experimental IETF draft. 654 4.5. The KSK Rollover Experiment in Yeti 656 The Yeti project provides a good basis to conduct a real-world 657 experiment of a root KSK roll. It is not a perfect analogy to the 658 IANA root because all of the resolvers to the Yeti experiment are 659 "opt-in", and are presumably run by administrators who are interested 660 in the DNS and knowledgeable about it. Still, it can inform the IANA 661 root KSK roll. 663 The IANA root KSK has not been rolled. ICANN put together a design 664 team to analyze the problem and make recommendations. The design 665 team put together a proposal [ICANN-ROOT-ROLL]. Whether this 666 proposal or a different one is adopted, the Yeti project can use it 667 as a basis for an experimental KSK roll. The experiment may not be 668 identical, since the time-lines laid out in the current IANA plan are 669 very long, and the Yeti project would like to conduct the experiment 670 in a shorter time. 672 The Yeti project would also like to conduct an experiment to try 673 rolling the root KSK using a straightforward method, such as a 674 double-DS approach outlined in [RFC6781]. If this ends up being 675 adopted for the IANA root, then only a single Yeti experiment will 676 need to be conducted. 678 < > 692 5. Other Technical findings and bugs 694 Besides the experiments with specific goal and procedures, some 695 unexpected bugs have been reported. It is worthwhile to record them 696 as technical findings from Yeti DNS Project. Hopefully, these 697 experiences can share and help. 699 5.1. IPv6 fragments issue 701 There are two cases in Yeti testbed reported that some Yeti root 702 servers on VPS failed to pull the zone from Distribution Master via 703 AXFR/IXFR. Two facts have been revealed in both client side and 704 server side after trouble shooting. 706 One fact in client side is that some operation system on VPS can not 707 handle IPv6 fragments correctly which causes failure when they are 708 doing AXRF/IXFR in TCP. The bug covers several OS and one VM 709 platform (listed below). 711 +-----------------------+-----------------+ 712 | OS | VM | 713 +-----------------------+-----------------+ 714 | NetBSD 6.1 and 7.0RC1 | VMware ESXI 5.5 | 715 | FreeBSD10.0 | | 716 | Debian 3.2 | | 717 +-----------------------+-----------------+ 719 Another fact is from server side in which one TCP segment of AXRF/ 720 IXFR is fragmented in IP layer resulting in two fragmented packets. 721 This weird behavior has been documented IETF draft 722 [I-D.andrews-tcp-and-ipv6-use-minmtu]. It reports a situation that 723 man implementations of TCP running over IPv6 neglect to check the 724 IPV6_USE_MIN_MTU value when performing MSS negotiation and when 725 constructing a TCP segment. It will cause TCP MSS option set to 1440 726 bytes, but IP layer will limit the packet less than 1280 bytes and 727 fragment the packets which finally result two fragmented packets. 729 The latter is not a technical error though, but it will cause the 730 error in the former fact which deserves much attention in IPv6 731 operation when VPS is already widely used. 733 5.2. Root compression issue 735 [RFC1035]specifies DNS massage compression scheme which allows a 736 domain name in a message to be represented as either: 1) a sequence 737 of labels ending in a zero octet, 2) a pointer, 3) or a sequence of 738 labels ending with a pointer. It is designed to save more room of 739 DNS packet. 741 However in Yeti testbed, it is found that Knot 2.0 server compresses 742 even the root. It means in a DNS message the name of root (a zero 743 octet) is replaced by a pointer of 2 octets. As well, it is legal 744 but breaks some tools (Go DNS lib in this bug report) which does not 745 expect such name compression for root. Now both Knot and Go DNS lib 746 have fixed that bug. 748 6. IANA Considerations 750 This document requires no action from the IANA. 752 7. Acknowledgements 754 The editors fully acknowledge that this memo is based on works and 755 discussions with Paul Vixie and Akira Kato in Yeti DNS 756 project[Yeti-DNS-Project]. Thanks to Antonio Prado and Stephane 757 Bortzmeyer who helped to review the document and give many editing 758 suggestions 760 Acknowledgment to all the Yeti participant and volunteers who make 761 the experimental testbed become functional and workable. 763 8. References 765 [dnscap-bug-report] 766 Bortzmeyer, S., "pcaputils: IWBN to have an option to run 767 a program after file rotation in pcapdump", 2009, 768 . 771 [I-D.andrews-tcp-and-ipv6-use-minmtu] 772 Andrews, M., "TCP Fails To Respect IPV6_USE_MIN_MTU", 773 draft-andrews-tcp-and-ipv6-use-minmtu-04 (work in 774 progress), October 2015. 776 [I-D.muks-dns-message-fragments] 777 Sivaraman, M., Kerr, S., and D. Song, "DNS message 778 fragments", draft-muks-dns-message-fragments-00 (work in 779 progress), July 2015. 781 [I-D.wkumari-dnsop-trust-management] 782 Kumari, W., Huston, G., Hunt, E., and R. Arends, 783 "Signalling of DNS Security (DNSSEC) Trust Anchors", 784 draft-wkumari-dnsop-trust-management-01 (work in 785 progress), October 2015. 787 [ICANN-ROOT-ROLL] 788 "Root Zone KSK Rollover Plan", 2015, 789 . 792 [Renumbering-J-Root] 793 Wessels, D., "Thirteen Years of "Old J-Root"", 2015, 794 . 798 [RFC1034] Mockapetris, P., "Domain names - concepts and facilities", 799 STD 13, RFC 1034, DOI 10.17487/RFC1034, November 1987, 800 . 802 [RFC1035] Mockapetris, P., "Domain names - implementation and 803 specification", STD 13, RFC 1035, DOI 10.17487/RFC1035, 804 November 1987, . 806 [RFC2460] Deering, S. and R. Hinden, "Internet Protocol, Version 6 807 (IPv6) Specification", RFC 2460, DOI 10.17487/RFC2460, 808 December 1998, . 810 [RFC4986] Eland, H., Mundy, R., Crocker, S., and S. Krishnaswamy, 811 "Requirements Related to DNS Security (DNSSEC) Trust 812 Anchor Rollover", RFC 4986, DOI 10.17487/RFC4986, August 813 2007, . 815 [RFC5011] StJohns, M., "Automated Updates of DNS Security (DNSSEC) 816 Trust Anchors", STD 74, RFC 5011, DOI 10.17487/RFC5011, 817 September 2007, . 819 [RFC6781] Kolkman, O., Mekking, W., and R. Gieben, "DNSSEC 820 Operational Practices, Version 2", RFC 6781, 821 DOI 10.17487/RFC6781, December 2012, 822 . 824 [RFC7626] Bortzmeyer, S., "DNS Privacy Considerations", RFC 7626, 825 DOI 10.17487/RFC7626, August 2015, 826 . 828 [ROOT-FAQ] 829 Karrenberg, D., "DNS Root Name Server FAQ", 2007, 830 . 832 [Root-Zone-Database] 833 "Root Zone Database", 834 . 836 [Yeti-DNS-Project] 837 "Website of Yeti DNS Project", . 839 Authors' Addresses 841 Linjian Song 842 Beijing Internet Institute 843 2/F, Building 5, No.58 Jinghai Road, BDA 844 Beijing 100176 845 P. R. China 847 Email: songlinjian@gmail.com 848 URI: http://www.biigroup.com/ 850 Shane Kerr 851 Beijing Internet Institute 852 2/F, Building 5, No.58 Jinghai Road, BDA 853 Beijing 100176 854 CN 856 Email: shane@biigroup.cn 857 URI: http://www.biigroup.com/ 859 Dong Liu 860 Beijing Internet Institute 861 2508 Room, 25th Floor, Tower A, Time Fortune 862 Beijing 100028 863 P. R. China 865 Email: dliu@biigroup.com 866 URI: http://www.biigroup.com/