idnits 2.17.1 draft-housley-web-pki-problems-02.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 : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (November 2, 2015) is 3098 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- == Missing Reference: 'RFC2560' is mentioned on line 118, but not defined ** Obsolete undefined reference: RFC 2560 (Obsoleted by RFC 6960) -- Obsolete informational reference (is this intentional?): RFC 5246 (Obsoleted by RFC 8446) -- Obsolete informational reference (is this intentional?): RFC 6844 (Obsoleted by RFC 8659) -- Obsolete informational reference (is this intentional?): RFC 6961 (Obsoleted by RFC 8446) -- Obsolete informational reference (is this intentional?): RFC 6962 (Obsoleted by RFC 9162) == Outdated reference: A later version (-02) exists of draft-shore-tls-dnssec-chain-extension-01 Summary: 1 error (**), 0 flaws (~~), 3 warnings (==), 5 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Internet Architecture Board R. Housley 3 Internet-Draft Vigil Security 4 Intended status: Informational K. O'Donoghue 5 Expires: May 5, 2016 Internet Society 6 November 2, 2015 8 Problems with the Public Key Infrastructure (PKI) for the World Wide Web 9 draft-housley-web-pki-problems-02.txt 11 Abstract 13 This document describes the technical and non-technical problems with 14 the current Public Key Infrastructure (PKI) used for the World Wide 15 Web. Some potential technical improvements are considered, and some 16 non-technical approaches to improvements are discussed. 18 Status of This Memo 20 This Internet-Draft is submitted in full conformance with the 21 provisions of BCP 78 and BCP 79. 23 Internet-Drafts are working documents of the Internet Engineering 24 Task Force (IETF). Note that other groups may also distribute 25 working documents as Internet-Drafts. The list of current Internet- 26 Drafts is at http://datatracker.ietf.org/drafts/current/. 28 Internet-Drafts are draft documents valid for a maximum of six months 29 and may be updated, replaced, or obsoleted by other documents at any 30 time. It is inappropriate to use Internet-Drafts as reference 31 material or to cite them other than as "work in progress." 33 This Internet-Draft will expire on May 5, 2016. 35 Copyright Notice 37 Copyright (c) 2015 IETF Trust and the persons identified as the 38 document authors. All rights reserved. 40 This document is subject to BCP 78 and the IETF Trust's Legal 41 Provisions Relating to IETF Documents 42 (http://trustee.ietf.org/license-info) in effect on the date of 43 publication of this document. Please review these documents 44 carefully, as they describe your rights and restrictions with respect 45 to this document. Code Components extracted from this document must 46 include Simplified BSD License text as described in Section 4.e of 47 the Trust Legal Provisions and are provided without warranty as 48 described in the Simplified BSD License. 50 Table of Contents 52 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 53 2. Very Brief Description of the Web PKI . . . . . . . . . . . . 2 54 3. Technical Improvements to the Web PKI . . . . . . . . . . . . 3 55 3.1. Weak Cryptographic Algorithms or Short Public Keys . . . 3 56 3.2. Certificate Status Checking . . . . . . . . . . . . . . . 4 57 3.2.1. Short-lived Certificates . . . . . . . . . . . . . . 5 58 3.2.2. CRL Distribution Points . . . . . . . . . . . . . . . 5 59 3.2.3. Proprietary Revocation Checks . . . . . . . . . . . . 5 60 3.2.4. OCSP Stapling . . . . . . . . . . . . . . . . . . . . 5 61 3.3. Surprising Certificates . . . . . . . . . . . . . . . . . 6 62 3.3.1. Certificate Authority Authorization (CAA) . . . . . . 7 63 3.3.2. HTTP Public Key Pinning (HPKP) . . . . . . . . . . . 8 64 3.3.3. HTTP Strict Transport Security (HSTS) . . . . . . . . 8 65 3.3.4. DNS-Based Authentication of Named Entities (DANE) . . 9 66 3.3.5. Certificate Transparency . . . . . . . . . . . . . . 10 67 3.4. Automation for Server Administrators . . . . . . . . . . 10 68 4. Policy and Process Improvements to the Web PKI . . . . . . . 11 69 4.1. Determination of the Trusted Certificate Authorities . . 11 70 4.2. Governance Structures for the Web PKI . . . . . . . . . . 12 71 5. Other Considerations for Improving the Web PKI . . . . . . . 13 72 6. Security Considerations . . . . . . . . . . . . . . . . . . . 13 73 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 13 74 8. References . . . . . . . . . . . . . . . . . . . . . . . . . 13 75 8.1. Normative References . . . . . . . . . . . . . . . . . . 13 76 8.2. Informative References . . . . . . . . . . . . . . . . . 13 77 Appendix A. Acknowledgements . . . . . . . . . . . . . . . . . . 16 78 Appendix B. IAB Members at the Time of Approval . . . . . . . . 16 79 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 16 81 1. Introduction 83 There are many technical and non-technical problems with the current 84 Public Key Infrastructure (PKI) used for the World Wide Web. This 85 document describes these problems, considers some potential technical 86 improvements, and discusses some non-technical approaches to 87 improvements. 89 The Web PKI makes use of certificates as described in RFC 5280 90 [RFC5280]. These certificates are primarily used with Transport 91 Layer Security (TLS) RFC 5246 [RFC5246]. 93 2. Very Brief Description of the Web PKI 95 Certificates are specified in [RFC5280]. Certificates contain, among 96 other things, a subject name and a public key, and they are digitally 97 signed by the Certification Authority (CA). Certificate users 98 require confidence that the private key associated with the certified 99 public key is owned by the named subject. A certificate has a 100 limited valid lifetime. 102 The architectural model used in the Web PKI includes: 104 EE: End Entity -- the subject of a certificate -- certificates are 105 issued to Web Servers, and certificates are also issued to 106 clients that need mutual authentication. 108 CA: Certification Authority -- the issuer of a certificate -- 109 issues certificates for Web Servers and clients. 111 RA: Registration Authority -- an optional system to which a CA 112 delegates some management functions such as identity validation 113 or physical credential distribution. 115 CAs are responsible for indicating the revocation status of the 116 certificates that they issue throughout the lifetime of the 117 certificate. Revocation status information may be provided using the 118 Online Certificate Status Protocol (OCSP) [RFC2560], certificate 119 revocation lists (CRLs) [RFC5280], or some other mechanism. In 120 general, when revocation status information is provided using CRLs, 121 the CA is also the CRL issuer. However, a CA may delegate the 122 responsibility for issuing CRLs to a different entity. 124 The enrollment process used by a CA makes sure that the subject name 125 in the certificate is appropriate and that the subject actually holds 126 the private key. Proof of possession of the private key is often 127 accomplished through a challenge-response protocol. 129 3. Technical Improvements to the Web PKI 131 Over the years, many technical improvements have been made to the Web 132 PKI. This section discusses sever problems and the technical 133 problems that have been made to address them. This history sets the 134 stage for suggestions for additional improvements in other sections 135 of this document. 137 3.1. Weak Cryptographic Algorithms or Short Public Keys 139 Over the years, the digital signature algorithms, one-way hash 140 functions, and public key sizes that are considered strong have 141 changed. This is not a surprise. Cryptographic algorithms age; they 142 become weaker with time. As new cryptanalysis techniques are 143 developed and computing capabilities improve, the work factor to 144 break a particular cryptographic algorithm will reduce. For this 145 reason, the algorithms and key sizes used in the Web PKI need to 146 migrate over time. A reasonable choice of algorithm or key size 147 needs to be evaluated periodically, and a transition may be needed 148 before the expected lifetime expires. 150 The browser vendors have been trying to manage algorithm and key size 151 transitions, but a long-lived trust anchor or intermediate CA 152 certificate can have a huge number of subordinate certificates. So, 153 removing a one because it uses a weak cryptographic algorithm or a 154 short public key can have a significant impact. 156 As a result, some valid trust anchors and certificates contain 157 cryptographic algorithms after weakness has been discovered and 158 widely known. Similarly, valid trust anchors and certificates 159 contain public keys after computational resources available to 160 attackers have rendered them too weak. We have seen a very 161 successful migration away from certificates that use the MD2 or MD5 162 one-way hash functions. However, there are still a great number of 163 certificates that use SHA-1 and 1024-bit RSA public keys, and these 164 should be replaced. 166 Today, the algorithms and key sizes used by a CA to sign certificates 167 with a traditional lifespan should offer 112 to 128 bits of security. 168 SHA-256 is a widely studied one-way hash function that meets this 169 requirement. RSA with a public key of at least 2048 bits or ECDSA 170 with a public key of at least 256 bits are widely studied digital 171 signature algorithms that meet this requirement. 173 3.2. Certificate Status Checking 175 Several years ago, many browsers do not perform certificate status 176 checks by default. That is, browsers did not check whether the 177 issuing CA has revoked the certificate unless the user explicitly 178 adjusted a setting to enable this feature. This check can be made by 179 fetching the most recent certificate revocation list (CRL) RFC 5280 180 [RFC5280], or this check can use the Online Certificate Status 181 Protocol (OCSP) RFC 6960 [RFC6960]. The location of the CRL or the 182 OCSP responder is usually found in the certificate itself. Either 183 one of these approaches add latency. The desire to provide a snappy 184 user experience is a significant reason that this feature was not 185 turned on by default. 187 Certificate status checking needs to be used at all times. Several 188 techniques have been tried by CAs and browsers to make certificate 189 status checking more efficient. Many CAs are using of Content 190 Delivery Networks (CDNs) by CAs to deliver CRLs and OCSP responses, 191 resulting in very high availability and low latency. Yet, browser 192 vendors are still reluctant to perform standard-based status checking 193 by default for every session. 195 3.2.1. Short-lived Certificates 197 Short-lived certificates are an excellent way to reduce the need for 198 certificate status checking. The shorter the life of the 199 certificate, the less time there is for anything to go wrong. If the 200 lifetime is short enough, policy might allow certificate status 201 checking can be skipped altogether. In practice, implementation of 202 short-lived certificates requires automation to assist web server 203 administrators, which is a topic that is discussed elsewhere in this 204 document. 206 3.2.2. CRL Distribution Points 208 The certificate revocation list distribution point (CRLDP) 209 certificate extension RFC 5280 [RFC5280] allows a CA to control the 210 maximum size of the CRLs that they issue. This helps in two ways. 211 First, the amount of storage needed by the browser to cache CRLs is 212 reduced. Second, and more importantly, the amount of time it takes 213 to download a CRL can be scoped, so that the amount of time needed to 214 fetch any single CRL is reasonable. 216 Few CAs take advantage of the CRLDP certificate extension to limit 217 the size of CRLs. In fact, there are several CAs that publish 218 extremely large CRLs. Browsers never want to suffer the latency 219 associated with large CRLs, and some ignore the CRLDP extension when 220 it is present. Browsers tend to avoid the use of CRLs altogether. 222 3.2.3. Proprietary Revocation Checks 224 Some browser vendors provide a proprietary mechanism for revocation 225 checking. These mechanisms obtain revocation status information once 226 per day for the entire Web PKI in a very compact form. No network 227 traffic is generated at the time that a certificate is being 228 validated, so there is no latency associated with revocation status 229 checking. The browser vendor infrastructure performs daily checks of 230 the Web PKI, and then the results are assembled in a proprietary 231 format and made available to the browser. These checks only cover 232 the trust anchor store for that browser vendor, so any trust anchors 233 added by the user cannot be checked in this manner. 235 3.2.4. OCSP Stapling 237 Browsers can avoid transmission of CRLs altogether by using the 238 Online Certificate Status Protocol (OCSP) RFC 6960 [RFC6960] to check 239 the validity of web server certificates. The TLS Certificate Status 240 Request extension is defined in Section 8 of RFC 6066 [RFC6066]. In 241 addition, RFC 6961 [RFC6961] defines the TLS Multiple Certificate 242 Status Request extension, which allows the web server to provide 243 status information about its own certificate and also the status of 244 intermediate certificates in the certification path. By including 245 this extension in the TLS handshake, the browser asks the web server 246 to provide an OCSP response in addition to its certificate. This 247 approach greatly reduces the number of round trips by the browser to 248 check the status of each certificate in the path. In addition, the 249 web server can cache the OCSP response for a period of time, avoiding 250 additional latency. Even in the cases where the web server needs to 251 contact the OCSP responder, the web server usually has a higher 252 bandwidth connection than the browser to do so. 254 The provision of the time-stamped OCSP response in the TLS handshake 255 is referred to as "stapling" the OCSP response to the TLS handshake. 256 If the browser does not receive a stapled OCSP response, it can 257 contact the OCSP responder directly. In addition, the MUST_STAPLE 258 feature [TLSFEATURE] can be used to insist that OCSP stapling be 259 used. 261 When every browser that connects to a high volume website performs 262 its own OCSP lookup, the OCSP responder must handle a real-time 263 response to every browser. OCSP stapling can avoid enormous volumes 264 of OCSP requests for certificates of popular websites, so stapling 265 can significantly reduce the cost of providing an OCSP service. 267 OCSP stapling can also improve user privacy, since the web server, 268 not the browser, contacts the OCSP responder. In this way, the OCSP 269 responder is not able to determine which browsers are checking the 270 validity of certificate for websites. 272 Many web site are taking advantage of OCSP sampling. At the time of 273 this writing, browser venders report that about 12% the the 274 transactions use OCSP sampling, and the number is on the rise. 276 3.3. Surprising Certificates 278 All of the CAs in the trust store are equally trusted for the entire 279 domain name space, so any CA can issue for any domain name. In fact, 280 there have been certificates issued by CAs that are surprising to the 281 legitimate owner of a domain. The domain name owner is surprised 282 because they did not request the certificates. They are initially 283 unaware that a CA has issued a certificate that contains their domain 284 name, and once the surprising certificate is discovered, it can be 285 very difficult for the legitimate domain name owner to get it 286 revoked. Further, browsers and other relying parties cannot 287 distinguish a certificate that the legitimate domain name owner 288 requested from an surprising one. 290 Since all of the CAs in the trust store are equally trusted, any CA 291 can issue a certificate for any domain name. There are known cases 292 where attackers have thwarted the CA protections and issued 293 certificates that were then used to mount attacks against the users 294 of that web site [FOXIT]. For this reason, all of the CAs listed in 295 the trust store must be very well protected. 297 The Baseline Requirements produced by the CA/Browser Forum [CAB2014] 298 tell CAs the checks that need to be performed before a certificate is 299 issued. In addition, WebTrust [WEBTRUST] provides the audit 300 requirements for CAs, and browser vendors will remove a CA from the 301 trust anchor store if successful audit reports are not made 302 available. 304 When a CA issues a certificate to a subordinate CA, the inclusion of 305 widely supported certificate extensions can reduce set of privileges 306 given to the sub-CA. This aligns with the traditional security 307 practice of least privilege, where only the privileges needed to 308 perform the envisioned tasks are provided. However, many sub-CAs 309 have certificates that pass along the full powers of the CA, creating 310 additional high-pay-off targets for attackers, and these sub-CAs may 311 not be held to the same certificate issuance requirements and audit 312 requirement as the parent CA. 314 Some major implementations have not fully implemented the mechanisms 315 necessary to reduce sub-CA privileges. For example, RFC 5280 316 [RFC5280] includes the specification of name constraints, and the CA/ 317 Browser Forum guidelines [CAB2014] encourage the use dNSNames in 318 permittedSubtrees within the name Constraints extension. Despite 319 this situation, one major browser does not support name constraints, 320 and as a result, CAs are reluctant to use them. Further, global CAs 321 are prepared to issue certificates within every top-level domain, 322 including ones that are newly-approved. It is not practical for 323 these global CAs to use name constraints in their sub-CA 324 certificates. 326 As a result of procedural failures or attacks, surprising 327 certificates are being issued. Several mechanisms have been defined 328 to avoid the issuance of surprising certificates or prevent browsers 329 from accepting them. 331 3.3.1. Certificate Authority Authorization (CAA) 333 The Certificate Authority Authorization (CAA) [RFC6844] DNS resource 334 record allows a domain administrator to specify one or more CA that 335 is authorized to issue certificates that include the domain name. 336 Then, a trustworthy CA will refuse to issue a certificate for a 337 domain name that has a CAA resource record that does not explicitly 338 name the CA. 340 To date, only one major CA performs this check, and there is no 341 indication that other CAs are planning to add this check in the near 342 future. 344 3.3.2. HTTP Public Key Pinning (HPKP) 346 HTTP Public Key Pinning (HPKP) [RFC7469] allows a web server to 347 instruct browsers to remember the server's public key fingerprints 348 for a period of time. The fingerprint is a one-way hash of the 349 subject public key information in the certificate. The Public-Key- 350 Pins header provides a maximum retention period, fingerprints of the 351 web server certificate, and optionally fingerprints for backup 352 certificates. The act of saving of the fingerprints is referred to 353 as "pinning". During pin lifetime, browsers require that the same 354 web server present a certificate chain that includes a public key 355 that matches one of the retained fingerprints. This prevents 356 impersonation of the website with a surprising certificate. 358 A website can choose to pin the CA certificate so that the browser 359 will accept only valid certificates for the website domain that are 360 issued by that CA. Alternatively, the website can choose to pin 361 their own certificate and at least one backup certificate in case the 362 current certificate needs to be replaced due to a compromised server. 364 Some browser vendors also pin certificates by hardcoding fingerprints 365 of very well known websites. 367 When HPKP is used, browsers may be able to detect a man-in-the- 368 middle. Sometimes the man-in-the-middle is an attacker, and other 369 times a service provider purposefully terminates the TLS at a 370 location other than the web server. One example became very public 371 in February 2012 when Trustwave admitted that it had issued a 372 subordinate CA certificate for use by a company to inspect corporate 373 network traffic [LC2012]. When HPKP is used, the browser user will 374 be notified if the key-pining is violated, unless the violating 375 certificate can be validated to a locally installed trust anchor. In 376 this situation, the browser is assuming that the user intended to 377 explicitly trust the certificate. 379 3.3.3. HTTP Strict Transport Security (HSTS) 381 HTTP Strict Transport Security (HSTS) [RFC6797] is a security policy 382 mechanism that protects secure websites against downgrade attacks, 383 and it greatly simplifies protection against cookie hijacking. The 384 presence of the Strict-Transport-Security header tells browsers that 385 all interactions with this web server should never use HTTP without 386 TLS, providing protection against eavesdropping and active network 387 attacks. 389 When a web server includes the Strict-Transport-Security header, the 390 browser is expected to do two things. First, the browser 391 automatically turns any insecure links into secure ones. For 392 instance, "http://mysite.example.com/mypage/" will be changed to 393 "https://mysite.example.com/mypage/". Second, if the TLS Handshake 394 results in some failure, such as the certificate cannot be validated, 395 then an error message is displayed and the user is denied access the 396 web application. 398 3.3.4. DNS-Based Authentication of Named Entities (DANE) 400 The DNS-Based Authentication of Named Entities (DANE) [RFC6698] 401 allows domain administrators to specify the raw public keys or 402 certificates that are used by web servers in their domain. DANE 403 leverages the DNS Security Extensions (DNSSEC) [RFC4034][RFC4035], 404 which provides digital signatures over DNS zones that are validated 405 with keys that are bound to the domain name of the signed zone. The 406 keys associated with a domain name can only be signed by a key 407 associated with the parent of that domain name. For example, the 408 DNSSEC keys for "www.example.com" can only be signed by the DNSSEC 409 keys for "example.com". Therefore, a malicious actor can only 410 compromise the keys of their own subdomains. Like the Web PKI, 411 DNSSEC relies on public keys used to validate chains of signatures, 412 but DNSSEC has a single root domain as opposed to a multiplicity of 413 trusted CAs. 415 DANE binds raw public keys or certificates to DNS names. The domain 416 administrator is the one that vouches for the binding of the public 417 key or the certificate to the domain name by adding the TSLA records 418 to the zone and then signing the zone. In this way, the same 419 administrator is responsible for managing the DNS names themselves 420 and associated public keys or certificates with those names. DANE 421 restricts the scope of assertions that can be made, forcing them to 422 be consistent with the DNS naming hierarchy. 424 In addition, DNSSEC reduces opportunities for redirection attacks by 425 binding the domain name to the public key or certificate. 427 Some Web PKI certificates are being posted in TLSA records, but 428 browsers expect to receive the the server certificate in the TLS 429 handshake, and there is little incentive to confirm that the received 430 certificate matches the one posted in the DNS. For this reason, work 431 has begun on a TLS extension that will allow the DNSSEC-protected 432 information to be provided in the handshake, which will eliminate the 433 latency [TLSCHAIN]. 435 3.3.5. Certificate Transparency 437 Certificate Transparency (CT) [RFC6962] offers a mechanism to detect 438 mis-issued certificates, and once detected, administrators and CAs 439 can take the necessary actions to revoke the mis-issued certificates. 441 When requesting a certificate, the administrator can request the CA 442 to include an embedded Signed Certificate Timestamp (SCT) in the 443 certificate to ensure that their legitimate certificate is logged 444 with one or more CT log. 446 An administrator, or another party acting on behalf of the 447 administrator, is able to monitor one or more CT log to which a pre- 448 certificate or certificate is submitted, and detect the logging of a 449 pre-certificate or certificate that contains their domain name. When 450 such a pre-certificate or certificate is detected, the CA can be 451 contacted to to get the mis-issued certificate revoked. 453 In the future, a browser may choose to reject certificates that do 454 not contain an SCT, and potentially notify the website administrator 455 or CA when they encounter such a certificate. Such reporting will 456 help detect mis-issuance of certificates and lead to their 457 revocation. 459 3.4. Automation for Server Administrators 461 There have been several attempts to provide automation for routine 462 tasks that are performed by web server administrators, such as 463 certificate renewal. For example, some commercial tools offer 464 automated certificate renewal and installation [DCEI][SSLM]. Also, 465 at least one proposal was brought to the IETF that allows a web 466 server automate obtaining and renewing certificates [PHBOB]. Without 467 automation, there are many manual steps involved in getting a 468 certificate from a CA, and to date none of these attempts at 469 automation have not enjoyed widespread interoperability and adoption. 470 There are at least two ways that this impacts web security. First, 471 many web sites do not have a certificate at all. The cost, time, and 472 effort are too great for the system administrator to go through the 473 effort, especially if the web site does not offer anything for 474 purchase. Second, once a certificate is obtained, a replacement is 475 not obtained until the current one expires. Automation can reduce 476 the amount of time that an administrator needs to dedicate to 477 certificate management, and it can make certificate renewal timely 478 and automatic. 480 The IETF ACME working group [ACMEWG] is working on protocols that 481 will provide system administrators an automated way to enroll and 482 renew their certificates. The expectation is that these 483 specifications will lead to widely available and interoperable tools 484 for system administrators. The expectation is that these protocols 485 and tools will be supported by all web server environments and CAs, 486 which will greatly reduce complexity and cost. 488 4. Policy and Process Improvements to the Web PKI 490 As with many technologies, the issues and complexities associated 491 with Web PKI use and deployment are just as much policy and process 492 as technical. These have evolved over time as well. This section 493 discusses the ways that business models and operational policies and 494 processes impact the Web PKI. 496 4.1. Determination of the Trusted Certificate Authorities 498 A very basic question for users of the Web PKI is "Who do you trust?" 499 The system for determining which CAs are added to or removed from the 500 trust store in browsers has been perceived by some as opaque and 501 confusing. As mentioned earlier, the CA/Browser Forum has developed 502 baseline requirements for the management and issuance of certificates 503 [CAB2014] for individual CAs. However, the process by which an 504 individual CA gets added to the trust store for each of the major 505 browsers is not straightforward. The individual browser vendors 506 determine what should and should not be trusted by including those 507 trusted CAs in their trust store. They do this by leveraging the 508 AICPA/CICA WebTrust Program for Certification Authorities [WEBTRUST]. 509 This program provides auditing requirements and a trust mark for CAs. 510 Failure to pass an audit can result in the CA being removed from the 511 trust store. 513 Once the browser has shipped, how does a user know which CAs are 514 trusted or what has changed recently. For an informed user, 515 information about which CAs have been added to or deleted from the 516 browser trust store can be found in the release notes. Users can 517 also examine the policies of the various CAs which would have been 518 developed and posted for the WebTrust Program. However, this may be 519 considered a fairly high barrier for the average user. There are 520 also options to make local modifications by educated users, but there 521 is little understanding about the implications of these choices. How 522 does an individual, organization, or enterprise really determine if a 523 particular CA is trustworthy? Do the default choices inherited from 524 the browser vendors truly represent the organization's trust model? 525 What constitutes sufficiently bad behavior by a CA to cause removal 526 from the trust store? 527 One form of bad behavior by CAs is the mis-issuance of certificates. 528 This mis-issuance can be either an honest mistake by the CA, 529 malicious behavior by the CA, or a case where an external party has 530 duped the CA into the mis-issuance. When a CA has delegated 531 authority to a sub-CA, and then the sub-CA issued bad certificates 532 either unintentionally or maliciously, the CA is able to deny 533 responsibility for the actions of the sub-CA. However, the CA may be 534 the only party that can revoke the sub-CA certificate to protect the 535 overall Web PKI. 537 Another complication with CAs and the trust store maintained by the 538 browser vendor is an enterprise managed PKI. For example, the US 539 Department of Defense operates its own PKI. In this case, the 540 enterprise maintains its own PKI for the exclusive use of the 541 enterprise itself. A bridge CA may be used to connect related 542 enterprises. The complication in this approach is that the 543 revocation mechanisms don't work with any additions that have been 544 made by the enterprise. See Section 3.2.3 on proprietary revocation 545 checks. 547 What constitutes sufficiently bad behavior by a CA to cause removal 548 from the trust store? The guidelines provided by the WebTrust 549 program [WEBTRUST] provide a framework, but the implications of 550 removing a CA can be significant. There may be a few very large CAs 551 that are critical to significant portions of Internet infrastructure. 552 Removing one of these trusted CAs can have a significant impact on a 553 large cross section of Internet users. 555 4.2. Governance Structures for the Web PKI 557 There are a number of organizations that play significant roles in 558 the operation of the Web PKI, including the CAB Forum, the WebTrust 559 Program, and the browser vendors. These organizations act on behalf 560 of the entire Internet community. Transparency in these operations 561 is vital to basic trust in the Web PKI. As one example, in the past 562 the CAB Forum was perceived as being a closed forum; however, some 563 changes were made to the operational procedures to allow more 564 visibility if not actual participation in the process [CAB1.2]. How 565 do we ensure that these processes continue to evolve in an open, 566 inclusive, and transparent manner? Currently, as the name implies, 567 the CAB Forum members represent CAs and browser vendors. How do we 568 ensure that relying parties a voice in this forum? 570 Since the Web PKI is widespread, applications beyond the World Wide 571 Web are making use of the Web PKI. For example, the Web PKI is used 572 to secure the connections between SMTP servers. In these 573 environments, the browser-centric capabilities are unavailable. For 574 example, see Section 3.2.3 on proprietary revocation checks. The 575 current governance structure does not provide a way for these other 576 applications to participate. How do we ensure that these other 577 applications get a voice in this forum? 579 5. Other Considerations for Improving the Web PKI 581 Other factors impact the usability and reliability of the Web PKI. 582 One factor is time synchronization. As time synchronization 583 infrastructure is made more secure, this infrastrucre will require 584 the use of certificates to authenticate time servers. However, 585 certificate infrastructure is reliant on quality time synchronization 586 as well, creating a boot strapping issue. 588 6. Security Considerations 590 Many people find browser error messages related to certificates 591 confusing. Good man-machine interfaces are always difficult, but in 592 this situation users are unable to understand the risks that they are 593 accepting by clicking "okay". This aspect of browser usability needs 594 to be improved for users to make better security choices. 596 7. IANA Considerations 598 None. 600 {{{ RFC Editor: Please remove this section prior to publication. }}} 602 8. References 604 8.1. Normative References 606 [RFC5280] Cooper, D., Santesson, S., Farrell, S., Boeyen, S., 607 Housley, R., and W. Polk, "Internet X.509 Public Key 608 Infrastructure Certificate and Certificate Revocation List 609 (CRL) Profile", RFC 5280, DOI 10.17487/RFC5280, May 2008, 610 . 612 8.2. Informative References 614 [ACMEWG] IETF, "Charter for Automated Certificate Management 615 Environment (acme) Working Group", June 2015, 616 . 618 [CAB1.2] CA/Browser Forum, "Bylaws of the CA/Browser Forum", 619 October 2014, . 622 [CAB2014] CA/Browser Forum, "CA/Browser Forum Baseline Requirements 623 for the Issuance and Management of Publicly-Trusted 624 Certificates, v.1.2.2", October 2014, 625 . 627 [DCEI] DigiCert Inc, "Express Install(TM): Automate SSL 628 Certificate Installation and HTTPS Configuration", AUGUST 629 2015, . 631 [FOXIT] Prins, J., "DigiNotar Certificate Authority breach: 632 "Operation Black Tulip"", September 2011, 633 . 638 [LC2012] Constantin, L., "Trustwave admits issuing man-in-the- 639 middle digital certificate; Mozilla debates punishment", 640 February 2012, 641 . 645 [PHBOB] Hallam-Baker, P., "OmniBroker Publication Protocol", 646 draft-hallambaker-omnipublish-00 (work in progress), May 647 2014. 649 [RFC4034] Arends, R., Austein, R., Larson, M., Massey, D., and S. 650 Rose, "Resource Records for the DNS Security Extensions", 651 RFC 4034, DOI 10.17487/RFC4034, March 2005, 652 . 654 [RFC4035] Arends, R., Austein, R., Larson, M., Massey, D., and S. 655 Rose, "Protocol Modifications for the DNS Security 656 Extensions", RFC 4035, DOI 10.17487/RFC4035, March 2005, 657 . 659 [RFC5246] Dierks, T. and E. Rescorla, "The Transport Layer Security 660 (TLS) Protocol Version 1.2", RFC 5246, 661 DOI 10.17487/RFC5246, August 2008, 662 . 664 [RFC6066] Eastlake 3rd, D., "Transport Layer Security (TLS) 665 Extensions: Extension Definitions", RFC 6066, 666 DOI 10.17487/RFC6066, January 2011, 667 . 669 [RFC6698] Hoffman, P. and J. Schlyter, "The DNS-Based Authentication 670 of Named Entities (DANE) Transport Layer Security (TLS) 671 Protocol: TLSA", RFC 6698, DOI 10.17487/RFC6698, August 672 2012, . 674 [RFC6797] Hodges, J., Jackson, C., and A. Barth, "HTTP Strict 675 Transport Security (HSTS)", RFC 6797, 676 DOI 10.17487/RFC6797, November 2012, 677 . 679 [RFC6844] Hallam-Baker, P. and R. Stradling, "DNS Certification 680 Authority Authorization (CAA) Resource Record", RFC 6844, 681 DOI 10.17487/RFC6844, January 2013, 682 . 684 [RFC6960] Santesson, S., Myers, M., Ankney, R., Malpani, A., 685 Galperin, S., and C. Adams, "X.509 Internet Public Key 686 Infrastructure Online Certificate Status Protocol - OCSP", 687 RFC 6960, DOI 10.17487/RFC6960, June 2013, 688 . 690 [RFC6961] Pettersen, Y., "The Transport Layer Security (TLS) 691 Multiple Certificate Status Request Extension", RFC 6961, 692 DOI 10.17487/RFC6961, June 2013, 693 . 695 [RFC6962] Laurie, B., Langley, A., and E. Kasper, "Certificate 696 Transparency", RFC 6962, DOI 10.17487/RFC6962, June 2013, 697 . 699 [RFC7469] Evans, C., Palmer, C., and R. Sleevi, "Public Key Pinning 700 Extension for HTTP", RFC 7469, DOI 10.17487/RFC7469, April 701 2015, . 703 [SSLM] Opsmate, Inc., "SSLMate: Secure your website the easy 704 way", August 2015, . 706 [TLSCHAIN] 707 Shore, M., Barnes, R., Huque, S., and W. Toorop, "X.509v3 708 TLS Feature Extension", draft-shore-tls-dnssec-chain- 709 extension-01 (work in progress), July 2015. 711 [TLSFEATURE] 712 Hallam-Baker, P., "X.509v3 TLS Feature Extension", draft- 713 hallambaker-tlsfeature-10 (work in progress), July 2015. 715 [WEBTRUST] 716 CPA Canada, "WebTrust Program for Certification 717 Authorities", August 2015, . 720 Appendix A. Acknowledgements 722 This document has been developed within the IAB Privacy and Security 723 Program. The authors greatly appreciate the review and suggestions 724 provided by Rick Andrews, Mary Barnes, Richard Barnes, Marc Blanchet, 725 Alissa Cooper, Nick Doty, Stephen Farrell, Joe Hall, Ted Hardie, 726 Ralph Holz, Christian Huitema, Eliot Lear, Xing Li, Lucy Lynch, 727 Gervase Markham, Andrei Robachevsky, Thomas Roessler, Jeremy Rowley, 728 Christine Runnegar, Jakob Schlyter, Wendy Seltzer, Brian Trammell, 729 and Juan Carlos Zuniga. 731 Appendix B. IAB Members at the Time of Approval 733 {{{ RFC Editor: Please add the names to the IAB members at the time 734 that this document is put into the RFC Editor queue. }}} 736 Authors' Addresses 738 Russ Housley 739 Vigil Security 740 918 Spring Knoll Drive 741 Herndon, VA 20170 742 USA 744 Email: housley@vigilsec.com 746 Karen O'Donoghue 747 Internet Society 748 1775 Wiehle Ave #201 749 Reston, VA 20190 750 USA 752 Email: odonoghue@isoc.org