idnits 2.17.1 draft-grimes-tcpm-tcpsce-01.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 (4 November 2019) is 1634 days in the past. Is this intentional? Checking references for intended status: Experimental ---------------------------------------------------------------------------- == Outdated reference: A later version (-04) exists of draft-morton-tsvwg-sce-00 == Outdated reference: A later version (-28) exists of draft-ietf-tcpm-accurate-ecn-09 Summary: 0 errors (**), 0 flaws (~~), 3 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 TCP Maintenance and Minor Extensions R.W. Grimes 3 Internet-Draft P. Heist 4 Intended status: Experimental 4 November 2019 5 Expires: 7 May 2020 7 Some Congestion Experienced in TCP 8 draft-grimes-tcpm-tcpsce-01 10 Abstract 12 This memo classifies a TCP code point ESCE ("Echo Some Congestion 13 Experienced") for use in feedback of IP code point SCE ("Some 14 Congestion Experienced"). 16 Status of This Memo 18 This Internet-Draft is submitted in full conformance with the 19 provisions of BCP 78 and BCP 79. 21 Internet-Drafts are working documents of the Internet Engineering 22 Task Force (IETF). Note that other groups may also distribute 23 working documents as Internet-Drafts. The list of current Internet- 24 Drafts is at https://datatracker.ietf.org/drafts/current/. 26 Internet-Drafts are draft documents valid for a maximum of six months 27 and may be updated, replaced, or obsoleted by other documents at any 28 time. It is inappropriate to use Internet-Drafts as reference 29 material or to cite them other than as "work in progress." 31 This Internet-Draft will expire on 7 May 2020. 33 Copyright Notice 35 Copyright (c) 2019 IETF Trust and the persons identified as the 36 document authors. All rights reserved. 38 This document is subject to BCP 78 and the IETF Trust's Legal 39 Provisions Relating to IETF Documents (https://trustee.ietf.org/ 40 license-info) in effect on the date of publication of this document. 41 Please review these documents carefully, as they describe your rights 42 and restrictions with respect to this document. Code Components 43 extracted from this document must include Simplified BSD License text 44 as described in Section 4.e of the Trust Legal Provisions and are 45 provided without warranty as described in the Simplified BSD License. 47 Table of Contents 49 1. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 2 50 2. Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 51 3. Background . . . . . . . . . . . . . . . . . . . . . . . . . 3 52 4. TCP Receiver . . . . . . . . . . . . . . . . . . . . . . . . 3 53 4.1. Single ACK implementation . . . . . . . . . . . . . . . . 3 54 4.2. Simple Delayed ACK implementation . . . . . . . . . . . . 3 55 4.3. Dithered Delayed ACK implementation . . . . . . . . . . . 3 56 4.4. Advanced ACK implementation . . . . . . . . . . . . . . . 4 57 4.5. ACK Thinning . . . . . . . . . . . . . . . . . . . . . . 4 58 5. TCP Sender . . . . . . . . . . . . . . . . . . . . . . . . . 4 59 6. Related Work . . . . . . . . . . . . . . . . . . . . . . . . 4 60 6.1. More Accurate ECN Feedback in TCP . . . . . . . . . . . . 4 61 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 5 62 8. Security Considerations . . . . . . . . . . . . . . . . . . . 5 63 9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . 5 64 10. Normative References . . . . . . . . . . . . . . . . . . . . 5 65 11. Informative References . . . . . . . . . . . . . . . . . . . 6 66 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 6 68 1. Terminology 70 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 71 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 72 "OPTIONAL" in this document are to be interpreted as described in 73 [RFC2119] and [RFC8174] when, and only when, they appear in all 74 capitals, as shown here. 76 2. Introduction 78 This memo requests a TCP header codepoint for use as ESCE. 80 This memo limits its scope to the definition of the TCP codepoint 81 ESCE, with a few brief illustrations of how it may be used. 83 SCE provides early and proportional feedback to the CC (congestion 84 control) algorithms for transport protocols, including but not 85 limited to TCP. The [sce-repo] is a Linux kernel modified to support 86 SCE, including: 88 * Enhancements to Linux's [cake] (Common Applications Kept Enhanced) 89 AQM to support SCE signaling 91 * Modifications to the TCP receive path to reflect SCE signals back 92 to the sender 94 * The addition of three new TCP CC algorithms that modify the 95 originals to add SCE support: Reno-SCE, DCTCP-SCE and Cubic-SCE 96 (work in progress as of this writing) 98 3. Background 100 [I-D.morton-tsvwg-sce] defines the IP SCE codepoint. 102 4. TCP Receiver 104 The mechanism defined to feed back SCE signals to the sender 105 explicitly makes use of the ESCE ("Echo Some Congestion Experienced") 106 code point in the TCP header. 108 4.1. Single ACK implementation 110 Upon receipt of a packet an ACK is immediatly generated, the SCE 111 codepoint is copied into the ESCE codepoint of the ACK. This keeps 112 the count of bytes SCE marked or not marked properly reflected in the 113 ACK packet(s). This valid implementation has the downside of 114 increasing ACK traffic. This implementation is NOT RECOMMENDED, but 115 useful for experimental work. 117 4.2. Simple Delayed ACK implementation 119 Upon receipt of a packet without an SCE codepoint traditional delayed 120 ACK processing is performed. Upon receipt of a packet with an SCE 121 codepoint immediate ACK processing SHOULD be done, this allows some 122 delaying of ACK's, but creates earlier feedback of the congested 123 state. This has the negative effect of over signalling ESCE. 125 4.3. Dithered Delayed ACK implementation 127 Upon receipt of a packet the SCE codepoint is stored in the TCP 128 state. Multiple packets state may be stored. Upon generation of an 129 ACK, normal or delayed, the stored SCE state is used to set the state 130 of ESCE. If no SCE state is in the TCP state, then the ESCE code 131 point MUST NOT be set. If all of the packets to be ACKed have SCE 132 state set then the ESCE code point MUST be set in the ACK. If some 133 of the packets to be ACKed have SCE state set then some proportional 134 number of ACK packets SHOULD be sent with the ESCE code point set. 135 Though this may defer a ESCE congestion signal when there is not a 136 next packet for some time it is generally accepted that such sparse 137 flows are not the source of congestion and thus the delayed signal is 138 of low impact. The goal is to have the same number of bytes marked 139 with ESCE as arrived with SCE. 141 4.4. Advanced ACK implementation 143 The Advanced ACK implementation actually immediately flushes any 144 pending ACK's up to the _previous_ segment when the state of the SCE 145 marking _changes_, allowing consecutive packets with the same SCE 146 state to be coalesced by the normal delayed-ack logic. The ACK 147 volume is then inflated only slightly compared to an unmarked 148 connection, and may actually involve fewer acks than a connection 149 involving CE marks or losses, during which delayed acks are 150 temporarily disabled. 152 4.5. ACK Thinning 154 Ack thinning is something that has been considered, given that [cake] 155 includes an optional ack-filter which does thinning. We have, for 156 example, added consideration of the ESCE bit to Cake's ack-filter. 157 Mathematically, the most extreme errors possible in either direction, 158 due to ack thinning, are easily corrected during subsequent RTTs. 160 5. TCP Sender 162 The recommended response to each single segment marked with ESCE is 163 to reduce cwnd by an amortised 1/sqrt(cwnd) segments. If the growth 164 rate is greater than that provided by the Reno-linear algorithm - eg. 165 slow-start exponential or CUBIC polynomial - then the growth rate 166 SHOULD also be reduced. 168 Other responses, such as the 1/cwnd from DCTCP, are also acceptable 169 but may perform less well. 171 There are no changes to the response functions with respect to CE or 172 packet loss specificed by this draft, hence [RFC3168] and [RFC8511] 173 are still applicable 175 This is still an area of continued investigation. 177 6. Related Work 179 6.1. More Accurate ECN Feedback in TCP [I-D.ietf-tcpm-accurate-ecn] 181 AccECN replaces the [RFC3168] definition of the ECE and CWR bits (and 182 the former NS bit) with its own three-bit field. This new 183 interpretation is predicated on successfully negotiating AccECN, and 184 is not useful to SCE implementations because it provides no 185 information about any ECT(1) codepoints received, and SCE does not 186 need or use the extra information about CE marks that the three-bit 187 field does provide. Hence SCE may be considered mutually exclusive 188 with AccECN on any given connection. 190 AccECN supports a fallback to [RFC3168] style signalling during the 191 three-way handshake by recognising the normal requests and responses 192 of an [RFC3168] endpoint. SCE endpoints also exhibit [RFC3168] 193 behaviour during the handshake, so this mutual exclusivity occurs 194 naturally. There will therefore be no confusion on the wire between 195 the two experiments, even though SCE does not explicitly negotiate 196 its upgrade from plain [RFC3168] behaviour. 198 The latter is consistent with the (now historic) Nonce Sum 199 specification, which also did not explicitly negotiate support, and 200 used the same additional ECN codepoint and TCP header bit that SCE is 201 now requesting. 203 7. IANA Considerations 205 This document requests one of the reserved bits in the TCP header, 206 with the former TCP NS ("Nonce Sum") bit (bit 7) being suggested due 207 to similarities with its previous usage. [RFC8311] (section 3) 208 obsoletes the NS codepoint making it avaliable for use. 210 8. Security Considerations 212 There are no Security considerations. 214 9. Acknowledgements 216 TBD 218 10. Normative References 220 [I-D.morton-tsvwg-sce] 221 Morton, J. and R. Grimes, "The Some Congestion Experienced 222 ECN Codepoint", draft-morton-tsvwg-sce-00 (work in 223 progress), 2 July 2019, 224 . 227 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 228 Requirement Levels", BCP 14, RFC 2119, 229 DOI 10.17487/RFC2119, March 1997, 230 . 232 [RFC8174] Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC 233 2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174, 234 May 2017, . 236 [RFC8311] Black, D., "Relaxing Restrictions on Explicit Congestion 237 Notification (ECN) Experimentation", RFC 8311, 238 DOI 10.17487/RFC8311, January 2018, 239 . 241 11. Informative References 243 [cake] "Cake - Common Applications Kept Enhanced", November 2019, 244 . 246 [I-D.ietf-tcpm-accurate-ecn] 247 Briscoe, B., Kuehlewind, M., and R. Scheffenegger, "More 248 Accurate ECN Feedback in TCP", draft-ietf-tcpm-accurate- 249 ecn-09 (work in progress), 8 July 2019, 250 . 253 [RFC3168] Ramakrishnan, K., Floyd, S., and D. Black, "The Addition 254 of Explicit Congestion Notification (ECN) to IP", 255 RFC 3168, DOI 10.17487/RFC3168, September 2001, 256 . 258 [RFC8511] Khademi, N., Welzl, M., Armitage, G., and G. Fairhurst, 259 "TCP Alternative Backoff with ECN (ABE)", RFC 8511, 260 DOI 10.17487/RFC8511, December 2018, 261 . 263 [sce-repo] "Some Congestion Experienced Reference Implementation 264 GitHub Repository", November 2019, 265 . 267 Authors' Addresses 269 Rodney W. Grimes 270 Redacted 271 Portland, OR 97217 272 United States 274 Email: rgrimes@freebsd.org 276 Peter G. Heist 277 Redacted 278 463 11 Liberec 30 279 Czech Republic 281 Email: pete@heistp.net