idnits 2.17.1 draft-ietf-p2psip-self-tuning-13.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 (June 14, 2014) is 3596 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Obsolete normative reference: RFC 5245 (Obsoleted by RFC 8445, RFC 8839) ** Obsolete normative reference: RFC 5389 (Obsoleted by RFC 8489) Summary: 2 errors (**), 0 flaws (~~), 1 warning (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 P2PSIP Working Group J. Maenpaa 3 Internet-Draft G. Camarillo 4 Intended status: Standards Track Ericsson 5 Expires: December 16, 2014 June 14, 2014 7 Self-tuning Distributed Hash Table (DHT) for REsource LOcation And 8 Discovery (RELOAD) 9 draft-ietf-p2psip-self-tuning-13.txt 11 Abstract 13 REsource LOcation And Discovery (RELOAD) is a peer-to-peer (P2P) 14 signaling protocol that provides an overlay network service. Peers 15 in a RELOAD overlay network collectively run an overlay algorithm to 16 organize the overlay, and to store and retrieve data. This document 17 describes how the default topology plugin of RELOAD can be extended 18 to support self-tuning, that is, to adapt to changing operating 19 conditions such as churn and network size. 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 December 16, 2014. 38 Copyright Notice 40 Copyright (c) 2014 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. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 3 57 3. Introduction to Stabilization in DHTs . . . . . . . . . . . . 5 58 3.1. Reactive vs. Periodic Stabilization . . . . . . . . . . . 5 59 3.2. Configuring Periodic Stabilization . . . . . . . . . . . 6 60 3.3. Adaptive Stabilization . . . . . . . . . . . . . . . . . 7 61 4. Introduction to Chord . . . . . . . . . . . . . . . . . . . . 8 62 5. Extending Chord-reload to Support Self-tuning . . . . . . . . 9 63 5.1. Update Requests . . . . . . . . . . . . . . . . . . . . . 10 64 5.2. Neighbor Stabilization . . . . . . . . . . . . . . . . . 10 65 5.3. Finger Stabilization . . . . . . . . . . . . . . . . . . 11 66 5.4. Adjusting Finger Table Size . . . . . . . . . . . . . . . 11 67 5.5. Detecting Partitioning . . . . . . . . . . . . . . . . . 11 68 5.6. Leaving the Overlay . . . . . . . . . . . . . . . . . . . 12 69 6. Self-tuning Chord Parameters . . . . . . . . . . . . . . . . 12 70 6.1. Estimating Overlay Size . . . . . . . . . . . . . . . . . 12 71 6.2. Determining Routing Table Size . . . . . . . . . . . . . 13 72 6.3. Estimating Failure Rate . . . . . . . . . . . . . . . . . 13 73 6.3.1. Detecting Failures . . . . . . . . . . . . . . . . . 14 74 6.4. Estimating Join Rate . . . . . . . . . . . . . . . . . . 15 75 6.5. Estimate Sharing . . . . . . . . . . . . . . . . . . . . 15 76 6.6. Calculating the Stabilization Interval . . . . . . . . . 17 77 7. Overlay Configuration Document Extension . . . . . . . . . . 18 78 8. Security Considerations . . . . . . . . . . . . . . . . . . . 18 79 9. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 19 80 9.1. Message Extensions . . . . . . . . . . . . . . . . . . . 19 81 9.2. A New IETF XML Registry . . . . . . . . . . . . . . . . . 19 82 10. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 19 83 11. References . . . . . . . . . . . . . . . . . . . . . . . . . 19 84 11.1. Normative References . . . . . . . . . . . . . . . . . . 19 85 11.2. Informative References . . . . . . . . . . . . . . . . . 20 86 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . 22 88 1. Introduction 90 REsource LOcation And Discovery (RELOAD) [RFC6940] is a peer-to-peer 91 signaling protocol that can be used to maintain an overlay network, 92 and to store data in and retrieve data from the overlay. For 93 interoperability reasons, RELOAD specifies one overlay algorithm, 94 called chord-reload, that is mandatory to implement. This document 95 extends the chord-reload algorithm by introducing self-tuning 96 behavior. 98 Distributed Hash Table (DHT) based overlay networks are self- 99 organizing, scalable and reliable. However, these features come at a 100 cost: peers in the overlay network need to consume network bandwidth 101 to maintain routing state. Most DHTs use a periodic stabilization 102 routine to counter the undesirable effects of churn on routing. To 103 configure the parameters of a DHT, some characteristics such as churn 104 rate and network size need to be known in advance. These 105 characteristics are then used to configure the DHT in a static 106 fashion by using fixed values for parameters such as the size of the 107 successor set, size of the routing table, and rate of maintenance 108 messages. The problem with this approach is that it is not possible 109 to achieve a low failure rate and a low communication overhead by 110 using fixed parameters. Instead, a better approach is to allow the 111 system to take into account the evolution of network conditions and 112 adapt to them. This document extends the mandatory-to-implement 113 chord-reload algorithm by making it self-tuning. Two main advantages 114 of self-tuning are that users no longer need to tune every DHT 115 parameter correctly for a given operating environment and that the 116 system adapts to changing operating conditions. 118 The remainder of this document is structured as follows: Section 2 119 provides definitions of terms used in this document. Section 3 120 discusses alternative approaches to stabilization operations in DHTs, 121 including reactive stabilization, periodic stabilization, and 122 adaptive stabilization. Section 4 gives an introduction to the Chord 123 DHT algorithm. Section 5 describes how this document extends the 124 stabilization routine of the chord-reload algorithm. Section 6 125 describes how the stabilization rate and routing table size are 126 calculated in an adaptive fashion. 128 2. Terminology 130 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 131 "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and 132 "OPTIONAL" in this document are to be interpreted as described in RFC 133 2119 [RFC2119]. 135 This document uses terminology and definitions from the RELOAD base 136 specification [RFC6940]. 138 numBitsInNodeId: Specifies the number of bits in a RELOAD Node-ID. 140 DHT: Distributed Hash Tables (DHTs) are a class of decentralized 141 distributed systems that provide a lookup service similar to a 142 regular hash table. Given a key, any peer participating in the 143 system can retrieve the value associated with that key. The 144 responsibility for maintaining the mapping from keys to values is 145 distributed among the peers. 147 Chord Ring: The Chord DHT uses ring topology and orders identifiers 148 on an identifier circle of size 2^numBitsInNodeId. This 149 identifier circle is called the Chord ring. On the Chord ring, 150 the responsibility for a key k is assigned to the node whose 151 identifier equals to or immediately follows k. 153 Finger Table: A data structure with up to (but typically less than) 154 numBitsInNodeId entries maintained by each peer in a Chord-based 155 overlay. The ith entry in the finger table of peer n contains the 156 identity of the first peer that succeeds n by at least 157 2^(numBitsInNodeId-i) on the Chord ring. This peer is called the 158 ith finger of peer n. As an example, the first entry in the 159 finger table of peer n contains a peer half-way around the Chord 160 ring from peer n. The purpose of the finger table is to 161 accelerate lookups. 163 n.id: An abbreviation that is in this document used refer to the 164 Node-ID of peer n. 166 O(g(n)): Informally, saying that some equation f(n) = O(g(n)) means 167 that f(n) is less than some constant multiple of g(n). For the 168 formal definition, please refer to [weiss1998]. 170 Omega(g(n)): Informally, saying that some equation f(n) = 171 Omega(g(n)) means that f(n) is more than some constant multiple of 172 g(n). For the formal definition, please refer to [weiss1998] 174 Percentile: The Pth (0<=P<=100) percentile of N values arranged in 175 ascending order is obtained by first calculating the (ordinal) 176 rank n=(P/100)*N, rounding the result to the nearest integer, and 177 then taking the value corresponding to that rank. 179 Predecessor List: A data structure containing the first r 180 predecessors of a peer on the Chord ring. 182 Successor List: A data structure containing the first r successors 183 of a peer on the Chord ring. 185 Neighborhood Set: A term used to refer to the set of peers included 186 in the successor and predecessor lists of a given peer. 188 Routing Table: Contents of a given peer's routing table include the 189 set of peers that the peer can use to route overlay messages. The 190 routing table is made up of the finger table, successor list and 191 predecessor list. 193 3. Introduction to Stabilization in DHTs 195 DHTs use stabilization routines to counter the undesirable effects of 196 churn on routing. The purpose of stabilization is to keep the 197 routing information of each peer in the overlay consistent with the 198 constantly changing overlay topology. There are two alternative 199 approaches to stabilization: periodic and reactive [rhea2004]. 200 Periodic stabilization can either use a fixed stabilization rate or 201 calculate the stabilization rate in an adaptive fashion. 203 3.1. Reactive vs. Periodic Stabilization 205 In reactive stabilization, a peer reacts to the loss of a peer in its 206 neighborhood set or to the appearance of a new peer that should be 207 added to its neighborhood set by sending a copy of its neighbor table 208 to all peers in the neighborhood set. Periodic recovery, in 209 contrast, takes place independently of changes in the neighborhood 210 set. In periodic recovery, a peer periodically shares its 211 neighborhood set with each or a subset of the members of that set. 213 The chord-reload algorithm [RFC6940] supports both reactive and 214 periodic stabilization. It has been shown in [rhea2004] that 215 reactive stabilization works well for small neighborhood sets (i.e., 216 small overlays) and moderate churn. However, in large-scale (e.g., 217 1000 peers or more [rhea2004]) or high-churn overlays, reactive 218 stabilization runs the risk of creating a positive feedback cycle, 219 which can eventually result in congestion collapse. In [rhea2004], 220 it is shown that a 1000-peer overlay under churn uses significantly 221 less bandwidth and has lower latencies when periodic stabilization is 222 used than when reactive stabilization is used. Although in the 223 experiments carried out in [rhea2004], reactive stabilization 224 performed well when there was no churn, its bandwidth use was 225 observed to jump dramatically under churn. At higher churn rates and 226 larger scale overlays, periodic stabilization uses less bandwidth and 227 the resulting lower contention for the network leads to lower 228 latencies. For this reason, most DHTs such as CAN [CAN], Chord 229 [Chord], Pastry [Pastry], Bamboo [rhea2004], etc. use periodic 230 stabilization [ghinita2006]. As an example, the first version of 231 Bamboo used reactive stabilization, which caused Bamboo to suffer 232 from degradation in performance under churn. To fix this problem, 233 Bamboo was modified to use periodic stabilization. 235 In Chord, periodic stabilization is typically done both for 236 successors and fingers. An alternative strategy is analyzed in 237 [krishnamurthy2008]. In this strategy, called the correction-on- 238 change maintenance strategy, a peer periodically stabilizes its 239 successors but does not do so for its fingers. Instead, finger 240 pointers are stabilized in a reactive fashion. The results obtained 241 in [krishnamurthy2008] imply that although the correction-on-change 242 strategy works well when churn is low, periodic stabilization 243 outperforms the correction-on-change strategy when churn is high. 245 3.2. Configuring Periodic Stabilization 247 When periodic stabilization is used, one faces the problem of 248 selecting an appropriate execution rate for the stabilization 249 procedure. If the execution rate of periodic stabilization is high, 250 changes in the system can be quickly detected, but at the 251 disadvantage of increased communication overhead. Alternatively, if 252 the stabilization rate is low and the churn rate is high, routing 253 tables become inaccurate and DHT performance deteriorates. Thus, the 254 problem is setting the parameters so that the overlay achieves the 255 desired reliability and performance even in challenging conditions, 256 such as under heavy churn. This naturally results in high cost 257 during periods when the churn level is lower than expected, or 258 alternatively, poor performance or even network partitioning in worse 259 than expected conditions. 261 In addition to selecting an appropriate stabilization interval, 262 regardless of whether periodic stabilization is used or not, an 263 appropriate size needs to be selected for the neighborhood set and 264 for the finger table. 266 The current approach is to configure overlays statically. This works 267 in situations where perfect information about the future is 268 available. In situations where the operating conditions of the 269 network are known in advance and remain static throughout the 270 lifetime of the system, it is possible to choose fixed optimal values 271 for parameters such as stabilization rate, neighborhood set size and 272 routing table size. However, if the operating conditions (e.g., the 273 size of the overlay and its churn rate) do not remain static but 274 evolve with time, it is not possible to achieve both a low lookup 275 failure rate and a low communication overhead by using fixed 276 parameters [ghinita2006]. 278 As an example, to configure the Chord DHT algorithm, one needs to 279 select values for the following parameters: size of successor list, 280 stabilization interval, and size of the finger table. To select an 281 appropriate value for the stabilization interval, one needs to know 282 the expected churn rate and overlay size. According to 283 [liben-nowell2002], a Chord network in a ring-like state remains in a 284 ring-like state as long as peers send Omega(square(log(N))) messages 285 before N new peers join or N/2 peers fail. Thus, in a 500-peer 286 overlay churning at a rate such that one peer joins and one peer 287 leaves the network every 30 seconds, an appropriate stabilization 288 interval would be on the order of 93s. According to [Chord], the 289 size of the successor list and finger table should be on the order of 290 log(N). Already a successor list of a modest size (e.g., log2(N) or 291 2*log2(N), which is the successor list size used in [Chord]) makes it 292 very unlikely that a peer will lose all of its successors, which 293 would cause the Chord ring to become disconnected. Thus, in a 294 500-peer network each peer should maintain on the order of nine 295 successors and fingers. However, if the churn rate doubles and the 296 network size remains unchanged, the stabilization rate should double 297 as well. That is, the appropriate maintenance interval would now be 298 on the order of 46s. On the other hand, if the churn rate becomes 299 e.g. six-fold and the size of the network grows to 2000 peers, on the 300 order of eleven fingers and successors should be maintained and the 301 stabilization interval should be on the order of 42s. If one 302 continued using the old values, this could result in inaccurate 303 routing tables, network partitioning, and deteriorating performance. 305 3.3. Adaptive Stabilization 307 A self-tuning DHT takes into consideration the continuous evolution 308 of network conditions and adapts to them. In a self-tuning DHT, each 309 peer collects statistical data about the network and dynamically 310 adjusts its stabilization rate, neighborhood set size, and finger 311 table size based on the analysis of the data [ghinita2006]. 312 Reference [mahajan2003] shows that by using self-tuning, it is 313 possible to achieve high reliability and performance even in adverse 314 conditions with low maintenance cost. Adaptive stabilization has 315 been shown to outperform periodic stabilization in terms of both 316 lookup failures and communication overhead [ghinita2006]. 318 4. Introduction to Chord 320 Chord [Chord] is a structured P2P algorithm that uses consistent 321 hashing to build a DHT out of several independent peers. Consistent 322 hashing assigns each peer and resource a fixed-length identifier. 323 Peers use SHA-1 as the base hash fuction to generate the identifiers. 324 As specified in RELOAD base, the length of the identifiers is 325 numBitsInNodeId=128 bits. The identifiers are ordered on an 326 identifier circle of size 2^numBitsInNodeId. On the identifier 327 circle, key k is assigned to the first peer whose identifier equals 328 or follows the identifier of k in the identifier space. The 329 identifier circle is called the Chord ring. 331 Different DHTs differ significantly in performance when bandwidth is 332 limited. It has been shown that when compared to other DHTs, the 333 advantages of Chord include that it uses bandwidth efficiently and 334 can achieve low lookup latencies at little cost [li2004]. 336 A simple lookup mechanism could be implemented on a Chord ring by 337 requiring each peer to only know how to contact its current successor 338 on the identifier circle. Queries for a given identifier could then 339 be passed around the circle via the successor pointers until they 340 encounter the first peer whose identifier is equal to or larger than 341 the desired identifier. Such a lookup scheme uses a number of 342 messages that grows linearly with the number of peers. To reduce the 343 cost of lookups, Chord maintains also additional routing information; 344 each peer n maintains a data structure with up to numBitsInNodeId 345 entries, called the finger table. The first entry in the finger 346 table of peer n contains the peer half-way around the ring from peer 347 n. The second entry contains the peer that is 1/4th of the way 348 around, the third entry the peer that is 1/8th of the way around, 349 etc. In other words, the ith entry in the finger table at peer n 350 contains the identity of the first peer s that succeeds n by at least 351 2^(numBitsInNodeId-i) on the Chord ring. This peer is called the ith 352 finger of peer n. The interval between two consecutive fingers is 353 called a finger interval. The ith finger interval of peer n covers 354 the range [n.id + 2^(numBitsInNodeId-i), n.id + 2^(numBitsInNodeId- 355 i+1)) on the Chord ring. In an N-peer network, each peer maintains 356 information about O(log(N)) other peers in its finger table. As an 357 example, if N=100000, it is sufficient to maintain 17 fingers. 359 Chord needs all peers' successor pointers to be up to date in order 360 to ensure that lookups produce correct results as the set of 361 participating peers changes. To achieve this, peers run a 362 stabilization protocol periodically in the background. The 363 stabilization protocol of the original Chord algorithm uses two 364 operations: successor stabilization and finger stabilization. 365 However, the Chord algorithm of RELOAD base defines two additional 366 stabilization components, as will be discussed below. 368 To increase robustness in the event of peer failures, each Chord peer 369 maintains a successor list of size r, containing the peer's first r 370 successors. The benefit of successor lists is that if each peer 371 fails independently with probability p, the probability that all r 372 successors fail simultaneously is only p^r. 374 The original Chord algorithm maintains only a single predecessor 375 pointer. However, multiple predecessor pointers (i.e., a predecessor 376 list) can be maintained to speed up recovery from predecessor 377 failures. The routing table of a peer consists of the successor 378 list, finger table, and predecessor list. 380 5. Extending Chord-reload to Support Self-tuning 382 This section describes how the mandatory-to-implement chord-reload 383 algorithm defined in RELOAD base [RFC6940] can be extended to support 384 self-tuning. 386 The chord-reload algorithm supports both reactive and periodic 387 recovery strategies. When the self-tuning mechanisms defined in this 388 document are used, the periodic recovery strategy MUST be used. 389 Further, chord-reload specifies that at least three predecessors and 390 three successors need to be maintained. When the self-tuning 391 mechanisms are used, the appropriate sizes of the successor list and 392 predecessor list are determined in an adaptive fashion based on the 393 estimated network size, as will be described in Section 6. 395 As specified in RELOAD base, each peer MUST maintain a stabilization 396 timer. When the stabilization timer fires, the peer MUST restart the 397 timer and carry out the overlay stabilization routine. Overlay 398 stabilization has four components in chord-reload: 400 1. Update the neighbor table. We refer to this as neighbor 401 stabilization. 403 2. Refreshing the finger table. We refer to this as finger 404 stabilization. 406 3. Adjusting finger table size. 408 4. Detecting partitioning. We refer to this as strong 409 stabilization. 411 As specified in RELOAD base [RFC6940], a peer sends periodic messages 412 as part of the neighbor stabilization, finger stabilization, and 413 strong stabilization routines. In neighbor stabilization, a peer 414 periodically sends an Update request to every peer in its Connection 415 Table. The default time is every ten minutes. In finger 416 stabilization, a peer periodically searches for new peers to include 417 in its finger table. This time defaults to one hour. This document 418 specifies how the neighbor stabilization and finger stabilization 419 intervals can be determined in an adaptive fashion based on the 420 operating conditions of the overlay. The subsections below describe 421 how this document extends the four components of stabilization. 423 5.1. Update Requests 425 As described in RELOAD base [RFC6940], the neighbor and finger 426 stabilization procedures are implemented using Update requests. 427 RELOAD base defines three types of Update requests: 'peer_ready', 428 'neighbors', and 'full'. Regardless of the type, all Update requests 429 include an 'uptime' field. Since the self-tuning extensions require 430 information on the uptimes of peers in the routing table, the sender 431 of an Update request MUST include its current uptime in seconds in 432 the 'uptime' field. 434 When self-tuning is used, each peer decides independently the 435 appropriate size for the successor list, predecessor list and finger 436 table. Thus, the 'predecessors', 'successors', and 'fingers' fields 437 included in RELOAD Update requests are of variable length. As 438 specified in RELOAD [RFC6940], variable length fields are on the wire 439 preceded by length bytes. In the case of the successor list, 440 predecessor list, and finger table, there are two length bytes 441 (allowing lengths up to 2^16-1). The number of NodeId structures 442 included in each field can be calculated based on the length bytes 443 since the size of a single NodeId structure is 16 bytes. If a peer 444 receives more entries than fit into its successor list, predecessor 445 list or finger table, the peer MUST ignore the extra entries. If a 446 peer receives less entries than it currently has in its own data 447 structure, the peer MUST NOT drop the extra entries from its data 448 structure. 450 5.2. Neighbor Stabilization 452 In the neighbor stabilization operation of chord-reload, a peer 453 periodically sends an Update request to every peer in its Connection 454 Table. In a small, low-churn overlay, the amount of traffic this 455 process generates is typically acceptable. However, in a large-scale 456 overlay churning at a moderate or high churn rate, the traffic load 457 may no longer be acceptable since the size of the connection table is 458 large and the stabilization interval relatively short. The self- 459 tuning mechanisms described in this document are especially designed 460 for overlays of the latter type. Therefore, when the self-tuning 461 mechanisms are used, each peer MUST send a periodic Update request 462 only to its first predecessor and first successor on the Chord ring. 464 The neighbor stabilization routine MUST be executed when the 465 stabilization timer fires. To begin the neighbor stabilization 466 routine, a peer MUST send an Update request to its first successor 467 and its first predecessor. The type of the Update request MUST be 468 'neighbors'. The Update request MUST include the successor and 469 predecessor lists of the sender. If a peer receiving such an Update 470 request learns from the predecessor and successor lists included in 471 the request that new peers can be included in its neighborhood set, 472 it MUST send Attach requests to the new peers. 474 After a new peer has been added to the predecessor or successor list, 475 an Update request of type 'peer_ready' MUST be sent to the new peer. 476 This allows the new peer to insert the sender into its neighborhood 477 set. 479 5.3. Finger Stabilization 481 Chord-reload specifies two alternative methods for searching for new 482 peers to the finger table. Both of the alternatives can be used with 483 the self-tuning extensions defined in this document. 485 Immediately after a new peer has been added to the finger table, a 486 Probe request MUST be sent to the new peer to fetch its uptime. The 487 requested_info field of the Probe request MUST be set to contain the 488 ProbeInformationType 'uptime' defined in RELOAD base [RFC6940]. 490 5.4. Adjusting Finger Table Size 492 The chord-reload algorithm defines how a peer can make sure that the 493 finger table is appropriately sized to allow for efficient routing. 494 Since the self-tuning mechanisms specified in this document produce a 495 network size estimate, this estimate can be directly used to 496 calculate the optimal size for the finger table. This mechanism MUST 497 be used instead of the one specified by chord-reload. A peer MUST 498 use the network size estimate to determine whether it needs to adjust 499 the size of its finger table each time when the stabilization timer 500 fires. The way this is done is explained in Section 6.2. 502 5.5. Detecting Partitioning 504 This document does not require any changes to the mechanism chord- 505 reload uses to detect network partitioning. 507 5.6. Leaving the Overlay 509 As specified in RELOAD base [RFC6940], a leaving peer SHOULD send a 510 Leave request to all members of its neighbor table prior to leaving 511 the overlay. The overlay_specific_data field MUST contain the 512 ChordLeaveData structure. The Leave requests that are sent to 513 successors MUST contain the predecessor list of the leaving peer. 514 The Leave requests that are sent to the predecessors MUST contain the 515 successor list of the leaving peer. If a given successor can 516 identify better predecessors than are already included in its 517 predecessor lists by investigating the predecessor list it receives 518 from the leaving peer, it MUST send Attach requests to them. 519 Similarly, if a given predecessor identifies better successors by 520 investigating the successor list it receives from the leaving peer, 521 it MUST send Attach requests to them. 523 6. Self-tuning Chord Parameters 525 This section specifies how to determine an appropriate stabilization 526 rate and routing table size in an adaptive fashion. The proposed 527 mechanism is based on [mahajan2003], [liben-nowell2002], and 528 [ghinita2006]. To calculate an appropriate stabilization rate, the 529 values of three parameters must be estimated: overlay size N, failure 530 rate U, and join rate L. To calculate an appropriate routing table 531 size, the estimated network size N can be used. Peers in the overlay 532 MUST re-calculate the values of the parameters to self-tune the 533 chord-reload algorithm at the end of each stabilization period before 534 re-starting the stabilization timer. 536 6.1. Estimating Overlay Size 538 Techniques for estimating the size of an overlay network have been 539 proposed for instance in [mahajan2003], [horowitz2003], 540 [kostoulas2005], [binzenhofer2006], and [ghinita2006]. In Chord, the 541 density of peer identifiers in the neighborhood set can be used to 542 produce an estimate of the size of the overlay, N [mahajan2003]. 543 Since peer identifiers are picked randomly with uniform probability 544 from the numBitsInNodeId-bit identifier space, the average distance 545 between peer identifiers in the successor set is 546 (2^numBitsInNodeId)/N. 548 To estimate the overlay network size, a peer MUST compute the average 549 inter-peer distance d between the successive peers starting from the 550 most distant predecessor and ending to the most distant successor in 551 the successor list. The estimated network size MUST be calculated 552 as: 554 2^numBitsInNodeId 555 N = ------------------- 556 d 558 This estimate has been found to be accurate within 15% of the real 559 network size [ghinita2006]. Of course, the size of the neighborhood 560 set affects the accuracy of the estimate. 562 During the join process, a joining peer fills its routing table by 563 sending a series of Ping and Attach requests, as specified in RELOAD 564 base [RFC6940]. Thus, a joining peer immediately has enough 565 information at its disposal to calculate an estimate of the network 566 size. 568 6.2. Determining Routing Table Size 570 As specified in RELOAD base, the finger table must contain at least 571 16 entries. When the self-tuning mechanisms are used, the size of 572 the finger table MUST be set to max(ceiling(log2(N)), 16) using the 573 estimated network size N. 575 The size of the successor list MUST be set to ceiling(log2(N)). An 576 implementation MAY place a lower limit on the size of the successor 577 list. As an example, the implementation might require the size of 578 the successor list to be always at least three. 580 A peer MAY choose to maintain a fixed-size predecessor list with only 581 three entries as specified in RELOAD base. However, it is 582 RECOMMENDED that a peer maintains ceiling(log2(N)) predecessors. 584 6.3. Estimating Failure Rate 586 A typical approach is to assume that peers join the overlay according 587 to a Poisson process with rate L and leave according to a Poisson 588 process with rate parameter U [mahajan2003]. The value of U can be 589 estimated using peer failures in the finger table and neighborhood 590 set [mahajan2003]. If peers fail with rate U, a peer with M unique 591 peer identifiers in its routing table should observe K failures in 592 time K/(M*U). Every peer in the overlay MUST maintain a history of 593 the last K failures. The current time MUST be inserted into the 594 history when the peer joins the overlay. The estimate of U MUST be 595 calculated as: 597 k 598 U = --------, 599 M * Tk 601 where M is the number of unique peer identifiers in the routing 602 table, Tk is the time between the first and the last failure in the 603 history, and k is the number of failures in the history. If k is 604 smaller than K, the estimate MUST be computed as if there was a 605 failure at the current time. It has been shown that an estimate 606 calculated in a similar manner is accurate within 17% of the real 607 value of U [ghinita2006]. 609 The size of the failure history K affects the accuracy of the 610 estimate of U. One can increase the accuracy by increasing K. 611 However, this has the side effect of decreasing responsiveness to 612 changes in the failure rate. On the other hand, a small history size 613 may cause a peer to overreact each time a new failure occurs. In 614 [ghinita2006], K is set to 25% of the routing table size. Use of 615 this value is RECOMMENDED. 617 6.3.1. Detecting Failures 619 A new failure MUST be inserted to the failure history in the 620 following cases: 622 1. A Leave request is received from a neigbhor. 624 2. A peer fails to reply to a Ping request sent in the situation 625 explained below. If no packets have been received on a 626 connection during the past 2*Tr seconds (where Tr is the 627 inactivity timer defined by ICE [RFC5245]), a RELOAD Ping request 628 MUST be sent to the remote peer. RELOAD mandates the use of STUN 629 [RFC5389] for keepalives. STUN keepalives take the form of STUN 630 Binding Indication transactions. As specified in ICE [RFC5245], 631 a peer sends a STUN Binding Indication if there has been no 632 packet sent on a connection for Tr seconds. Tr is configurable 633 and has a default of 15 seconds. Although STUN Binding 634 Indications do not generate a response, the fact that a peer has 635 failed can be learned from the lack of packets (Binding 636 Indications or application protocol packets) received from the 637 peer. If the remote peer fails to reply to the Ping request, the 638 sender MUST consider the remote peer to have failed. 640 As an alternative to relying on STUN keepalives to detect peer 641 failure, a peer could send additional, frequent RELOAD messages to 642 every peer in its Connection Table. These messages could be Update 643 requests, in which case they would serve two purposes: detecting peer 644 failure and stabilization. However, as the cost of this approach can 645 be very high in terms of bandwidth consumption and traffic load, 646 especially in large-scale overlays experiencing churn, its use is NOT 647 RECOMMENDED. 649 6.4. Estimating Join Rate 651 Reference [ghinita2006] proposes that a peer can estimate the join 652 rate based on the uptime of the peers in its routing table. An 653 increase in peer join rate will be reflected by a decrease in the 654 average age of peers in the routing table. Thus, each peer MUST 655 maintain an array of the ages of the peers in its routing table 656 sorted in increasing order. Using this information, an estimate of 657 the global peer join rate L MUST be calculated as: 659 N 660 L = ----------------------, 661 Ages[floor(rsize/2)] 663 where Ages is an array containing the ages of the peers in the 664 routing table sorted in increasing order and rsize is the size of the 665 routing table. It has been shown that the estimate obtained by using 666 this method is accurate within 22% of the real join rate 667 [ghinita2006]. Of course, the size of the routing table affects the 668 accuracy. 670 In order for this mechanism to work, peers need to exchange 671 information about the time they have been present in the overlay. 672 Peers receive the uptimes of their successors and predecessors during 673 the stabilization operations since all Update requests carry uptime 674 values. A joining peer learns the uptime of the admitting peer since 675 it receives an Update from the admitting peer during the join 676 procedure. Peers learn the uptimes of new fingers since they can 677 fetch the uptime using a Probe request after having attached to the 678 new finger. 680 6.5. Estimate Sharing 682 To improve the accuracy of network size, join rate, and leave rate 683 estimates, peers MUST share their estimates. When the stabilization 684 timer fires, a peer MUST select number-of-peers-to-probe random peers 685 from its finger table and send each of them a Probe request. The 686 targets of Probe requests are selected from the finger table rather 687 than from the neighbor table since neighbors are likely to make 688 similar errors when calculating their estimates. number-of-peers-to- 689 probe is a new element in the overlay configuration document. It is 690 defined in Section 7. Both the Probe request and the answer returned 691 by the target peer MUST contain a new message extension whose 692 MessageExtensionType is 'self_tuning_data'. This extension type is 693 defined in Section 9.1. The extension_contents field of the 694 MessageExtension structure MUST contain a SelfTuningData structure: 696 struct { 697 uint32 network_size; 698 uint32 join_rate; 699 uint32 leave_rate; 700 } SelfTuningData; 702 The contents of the SelfTuningData structure are as follows: 704 network_size 706 The latest network size estimate calculated by the sender. 708 join_rate 710 The latest join rate estimate calculated by the sender. 712 leave_rate 714 The latest leave rate estimate calculated by the sender. 716 The join and leave rates are expressed as joins or failures per 24 717 hours. As an example, if the global join rate estimate a peer has 718 calculated is 0.123 peers/s, it would include in the join_rate field 719 the ceiling of the value 10627.2 (24*60*60*0.123 = 10627.2), that is, 720 the value 10628. 722 The 'type' field of the MessageExtension structure MUST be set to 723 contain the value 'self_tuning_data'. The 'critical' field of the 724 structure MUST be set to False. 726 A peer MUST store all estimates it receives in Probe requests and 727 answers during a stabilization interval. When the stabilization 728 timer fires, the peer MUST calculate the estimates to be used during 729 the next stabilization interval by taking the 75th percentile (i.e., 730 third quartile) of a data set containing its own estimate and the 731 received estimates. 733 The default value for number-of-peers-to-probe is 4. This default 734 value is recommended to allow a peer to receive a sufficiently large 735 set of estimates from other peers. With a value of 4, a peer 736 receives four estimates in Probe answers. On the average, each peer 737 also receives four Probe requests each carrying an estimate. Thus, 738 on the average, each peer has nine estimates (including its own) that 739 it can use at the end of the stablization interval. A value smaller 740 than 4 is NOT RECOMMENDED to keep the number of received estimates 741 high enough. As an example, if the value were 2, there would be 742 peers in the overlay that would only receive two estimates during a 743 stabilization interval. Such peers would only have three estimates 744 available at the end of the interval, which may not be reliable 745 enough since even a single exceptionally high or low estimate can 746 have a large impact. 748 6.6. Calculating the Stabilization Interval 750 According to [liben-nowell2002], a Chord network in a ring-like state 751 remains in a ring-like state as long as peers send 752 Omega(square(log(N))) messages before N new peers join or N/2 peers 753 fail. We can use the estimate of peer failure rate, U, to calculate 754 the time Tf in which N/2 peers fail: 756 1 757 Tf = ------ 758 2*U 760 Based on this estimate, a stabilization interval Tstab-1 MUST be 761 calculated as: 763 Tf 764 Tstab-1 = ----------------- 765 square(log2(N)) 767 On the other hand, the estimated join rate L can be used to calculate 768 the time in which N new peers join the overlay. Based on the 769 estimate of L, a stabilization interval Tstab-2 MUST be calculated 770 as: 772 N 773 Tstab-2 = --------------------- 774 L * square(log2(N)) 776 Finally, the actual stabilization interval Tstab that MUST be used 777 can be obtained by taking the minimum of Tstab-1 and Tstab-2. 779 The results obtained in [maenpaa2009] indicate that making the 780 stabilization interval too small has the effect of making the overlay 781 less stable (e.g., in terms of detected loops and path failures). 782 Thus, a lower limit should be used for the stabilization period. 783 Based on the results in [maenpaa2009], a lower limit of 15s is 784 RECOMMENDED, since using a stabilization period smaller than this 785 will with a high probability cause too much traffic in the overlay. 787 7. Overlay Configuration Document Extension 789 This document extends the RELOAD overlay configuration document by 790 adding one new element, "number-of-peers-to-probe", inside each 791 "configuration" element. 793 self-tuning:number-of-peers-to-probe: The number of fingers to which 794 Probe requests are sent to obtain their network size, join rate, 795 and leave rate estimates. The default value is 4. 797 The Relax NG Grammar for this element is: 799 namespace self-tuning = "urn:ietf:params:xml:ns:p2p:self-tuning" 801 parameter &= element self-tuning:number-of-peers-to-probe { 802 xsd:unsignedInt }? 804 This namespace is added into the element in the 805 overlay configuration file. 807 8. Security Considerations 809 In the same way as malicious or compromised peers implementing the 810 RELOAD base protocol [RFC6940] can advertise false network metrics or 811 distribute false routing table information for instance in RELOAD 812 Update messages, malicious peers implementing this specification may 813 share false join rate, leave rate, and network size estimates. For 814 such attacks, the same security concerns apply as in the RELOAD base 815 specification. In addition, as long as the amount of malicious peers 816 in the overlay remains modest, the statistical mechanisms applied in 817 Section 6.5 (i.e., the use of 75th percentiles) to process the shared 818 estimates a peer obtains help ensure that estimates that are clearly 819 different from (i.e., larger or smaller than) other received 820 estimates will not significantly influence the process of adapting 821 the stabilization interval and routing table size. However, it 822 should be noted that if an attacker is able to impersonate a high 823 number of other peers in the overlay in strategic locations, it may 824 be able to send a high enough number of false estimates to a victim 825 and therefore influence the victim's choice of a stabilization 826 interval. 828 9. IANA Considerations 830 9.1. Message Extensions 832 This document introduces one additional extension to the "RELOAD 833 Extensions" Registry: 835 +------------------+-------+---------------+ 836 | Extension Name | Code | Specification | 837 +------------------+-------+---------------+ 838 | self_tuning_data | 0x3 | RFC-AAAA | 839 +------------------+-------+---------------+ 841 The contents of the extension are defined in Section 6.5. 843 Note to RFC Editor: please replace AAAA with the RFC number for this 844 specification. 846 9.2. A New IETF XML Registry 848 This document registers one new URI for the self-tuning namespace in 849 the "ns" subregistry of the IETF XML registry defined in [RFC3688]. 851 URI: urn:ietf:params:xml:ns:p2p:self-tuning 853 Registrant Contact: The IESG 855 XML: N/A, the requested URI is an XML namespace 857 10. Acknowledgments 859 The authors would like to thank Jani Hautakorpi for his contributions 860 to the document. The authors would also like to thank Carlos 861 Bernardos, Martin Durst, Alissa Cooper, Tobias Gondrom, and Barry 862 Leiba for their comments on the document. 864 11. References 866 11.1. Normative References 868 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 869 Requirement Levels", BCP 14, RFC 2119, March 1997. 871 [RFC5245] Rosenberg, J., "Interactive Connectivity Establishment 872 (ICE): A Protocol for Network Address Translator (NAT) 873 Traversal for Offer/Answer Protocols", RFC 5245, April 874 2010. 876 [RFC5389] Rosenberg, J., Mahy, R., Matthews, P., and D. Wing, 877 "Session Traversal Utilities for NAT (STUN)", RFC 5389, 878 October 2008. 880 [RFC6940] Jennings, C., Lowekamp, B., Rescorla, E., Baset, S., and 881 H. Schulzrinne, "REsource LOcation And Discovery (RELOAD) 882 Base Protocol", RFC 6940, January 2014. 884 11.2. Informative References 886 [CAN] Ratnasamy, S., Francis, P., Handley, M., Karp, R., and S. 887 Schenker, "A Scalable Content-Addressable Network", In 888 Proceedings of the 2001 Conference on Applications, 889 Technologies, Architectures and Protocols for Computer 890 Communications pp. 161-172, August 2001. 892 [Chord] Stoica, I., Morris, R., Liben-Nowell, D., Karger, D., 893 Kaashoek, M., Dabek, F., and H. Balakrishnan, "Chord: A 894 Scalable Peer-to-peer Lookup Service for Internet 895 Applications", IEEE/ACM Transactions on Networking Volume 896 11, Issue 1, pp. 17-32, February 2003. 898 [Pastry] Rowstron, A. and P. Druschel, "Pastry: Scalable, 899 Decentralized Object Location and Routing for Large-Scale 900 Peer-to-Peer Systems", In Proceedings of the IFIP/ACM 901 International Conference on Distribued Systems Platforms 902 pp. 329-350, November 2001. 904 [RFC3688] Mealling, M., "The IETF XML Registry", BCP 81, RFC 3688, 905 January 2004. 907 [binzenhofer2006] 908 Binzenhofer, A., Kunzmann, G., and R. Henjes, "A Scalable 909 Algorithm to Monitor Chord-Based P2P Systems at Runtime", 910 In Proceedings of the 20th IEEE International Parallel and 911 Distributed Processing Symposium (IPDPS) pp. 1-8, April 912 2006. 914 [ghinita2006] 915 Ghinita, G. and Y. Teo, "An Adaptive Stabilization 916 Framework for Distributed Hash Tables", In Proceedings of 917 the 20th IEEE International Parallel and Distributed 918 Processing Symposium (IPDPS) pp. 29-38, April 2006. 920 [horowitz2003] 921 Horowitz, K. and D. Malkhi, "Estimating Network Size from 922 Local Information", Information Processing Letters Volume 923 88, Issue 5, pp. 237-243, December 2003. 925 [kostoulas2005] 926 Kostoulas, D., Psaltoulis, D., Gupta, I., Birman, K., and 927 A. Demers, "Decentralized Schemes for Size Estimation in 928 Large and Dynamic Groups", In Proceedings of the 4th IEEE 929 International Symposium on Network Computing and 930 Applications pp. 41-48, July 2005. 932 [krishnamurthy2008] 933 Krishnamurthy, S., El-Ansary, S., Aurell, E., and S. 934 Haridi, "Comparing Maintenance Strategies for Overlays", 935 In Proceedings of the 16th Euromicro Conference on 936 Parallel, Distributed and Network-Based Processing pp. 937 473-482, February 2008. 939 [li2004] Li, J., Strinbling, J., Gil, T., Morris, R., and M. 940 Kaashoek, "Comparing the Performance of Distributed Hash 941 Tables Under Churn", Peer-to-Peer Systems III, volume 3279 942 of Lecture Notes in Computer Science Springer, pp. 87-99, 943 February 2005. 945 [liben-nowell2002] 946 Liben-Nowell, D., Balakrishnan, H., and D. Karger, 947 "Observations on the Dynamic Evolution of Peer-to-Peer 948 Networks", In Proceedings of the 1st International 949 Workshop on Peer-to-Peer Systems (IPTPS) pp. 22-33, March 950 2002. 952 [maenpaa2009] 953 Maenpaa, J. and G. Camarillo, "A Study on Maintenance 954 Operations in a Chord-Based Peer-to-Peer Session 955 Initiation Protocol Overlay Network", In Proceedings of 956 the 23rd IEEE International Parallel and Distributed 957 Processing Symposium (IPDPS) pp. 1-9, May 2009. 959 [mahajan2003] 960 Mahajan, R., Castro, M., and A. Rowstron, "Controlling the 961 Cost of Reliability in Peer-to-Peer Overlays", In 962 Proceedings of the 2nd International Workshop on Peer-to- 963 Peer Systems (IPTPS) pp. 21-32, February 2003. 965 [rhea2004] 966 Rhea, S., Geels, D., Roscoe, T., and J. Kubiatowicz, 967 "Handling Churn in a DHT", In Proceedings of the USENIX 968 Annual Technical Conference pp. 127-140, June 2004. 970 [weiss1998] 971 Weiss, M., "Data Structures and Algorithm Analysis in 972 C++", Addison-Wesley Longman Publishin Co., Inc. 2nd 973 Edition, ISBN:0201361221, 1998. 975 Authors' Addresses 977 Jouni Maenpaa 978 Ericsson 979 Hirsalantie 11 980 Jorvas 02420 981 Finland 983 Email: Jouni.Maenpaa@ericsson.com 985 Gonzalo Camarillo 986 Ericsson 987 Hirsalantie 11 988 Jorvas 02420 989 Finland 991 Email: Gonzalo.Camarillo@ericsson.com