idnits 2.17.1 draft-ietf-bmwg-ospfconv-applicability-03.txt: ** The Abstract section seems to be numbered Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** Looks like you're using RFC 2026 boilerplate. This must be updated to follow RFC 3978/3979, as updated by RFC 4748. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- ** The document seems to lack a 1id_guidelines paragraph about Internet-Drafts being working documents. ** The document seems to lack a 1id_guidelines paragraph about 6 months document validity. == No 'Intended status' indicated for this document; assuming Proposed Standard == The page length should not exceed 58 lines per page, but there was 10 longer pages, the longest (page 2) being 60 lines == It seems as if not all pages are separated by form feeds - found 0 form feeds but 11 pages Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack an Introduction section. ** The document seems to lack a Security Considerations section. ** The document seems to lack an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) ** The abstract seems to contain references ([BENCHMARK], [TERM]), which it shouldn't. Please replace those with straight textual mentions of the documents in question. Miscellaneous warnings: ---------------------------------------------------------------------------- -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (March 2003) is 7712 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) -- Missing reference section? 'BENCHMARK' on line 394 looks like a reference -- Missing reference section? 'TERM' on line 61 looks like a reference -- Missing reference section? 'INTERCONNECT' on line 403 looks like a reference -- Missing reference section? 'FIB-TERM' on line 407 looks like a reference Summary: 8 errors (**), 0 flaws (~~), 3 warnings (==), 6 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group Vishwas Manral 3 Internet Draft Netplane Systems 4 Russ White 5 Cisco Systems 6 Aman Shaikh 7 Expiration Date: September 2003 University of California 8 File Name: draft-ietf-bmwg-ospfconv-applicability-03.txt March 2003 10 Benchmarking Applicability for Basic OSPF Convergence 11 draft-ietf-bmwg-ospfconv-applicability-03.txt 13 1. Status of this Memo 15 This document is an Internet-Draft and is in full conformance with 16 all provisions of Section 10 of RFC2026. 18 Internet Drafts are working documents of the Internet Engineering 19 Task Force (IETF), its Areas, and its Working Groups. Note that other 20 groups may also distribute working documents as Internet Drafts. 22 Internet Drafts are draft documents valid for a maximum of six 23 months. Internet Drafts may be updated, replaced, or obsoleted by 24 other documents at any time. It is not appropriate to use Internet 25 Drafts as reference material or to cite them other than as a "working 26 draft" or "work in progress". 28 The list of current Internet-Drafts can be accessed at 29 http://www.ietf.org/ietf/1id-abstracts.txt 31 The list of Internet-Draft Shadow Directories can be accessed at 32 http://www.ietf.org/shadow.html. 34 2. Abstract 36 This draft describes the applicability of [BENCHMARK] and similar 37 work which may be done in the future. Refer to [TERM] for terminology 38 used in this draft and [BENCHMARK]. The draft defines the advantages 39 as well as limitations of using the method defined in [BENCHMARK], 40 besides describing the pitfalls to avoid during measurement. 42 3. Motivation 44 There is a growing interest in testing single router control plane 45 convergence for routing protocols, with many people looking at 46 testing methodologies which can provide information on how long it 47 takes for a network to converge after various network events occur. 48 It is important to consider the framework within which any given 49 convergence test is executed when attempting to apply the results of 50 the testing, since the framework can have a major impact on the 51 results. For instance, determining when a network is converged, what 52 parts of the router's operation are considered within the testing, 53 and other such things will have a major impact on what apparent 54 performance routing protocols provide. 56 This document describes in detail the various benefits and pitfalls 57 of tests described in [BENCHMARK]. It also explains how such 58 measurements can be useful for providers and the research community. 60 NOTE: The word convergence within this document refers to single 61 router contorl plane convergence [TERM]. 63 4. Advantages of Such Measurement 65 o To be able to compare the iterations of a protocol implemen- 66 tation. It is often useful to be able to compare the perfor- 67 mance of two iterations of a given implementation of a proto- 68 col to determine where improvements have been made and where 69 further improvements can be made. 71 o To understand, given a set parameters (network conditions), 72 how a particular implementation on a particular device is 73 going to perform. For instance, if you were trying to decide 74 the processing power (size of device) required in a certain 75 location within a network, you can emulate the conditions 76 which are going to exist at that point in the network and use 77 the test described to measure the perfomance of several dif- 78 ferent routers. The results of these tests can provide one 79 possible data point for an intelligent decision. 81 If the device being tested is to be deployed in a running 82 network, using routes taken from the network where the equip- 83 ment is to be deployed rather than some generated topology in 84 these tests will give results which are closer to the real 85 preformance of the device. Care should be taken to emulate or 86 take routes from the actual location in the network where the 87 device will be (or would be) deployed. For instance, one set 88 of routes may be taken from an abr, one set from an area 0 89 only router, various sets from stub area, another set from 90 various normal areas, etc. 92 o To measure the performance of an OSPF implementation in a 93 wide variety of scenarios. 95 o To be used as parameters in OSPF simulations by researchers. 96 It may some times be required for certain kinds of research 97 to measure the individual delays of each parameter within an 98 OSPF implementation. These delays can be measured using the 99 methods defined in [BENCHMARK]. 101 o To help optimize certain configurable parameters. It may some 102 times be helpful for operators to know the delay required for 103 individual tasks so as to optimize the resource usage in the 104 network i.e. if it is found that the processing time is x 105 seconds on an router, it would be helpful to determine the 106 rate at which to flood LSA's to that router so as to not 107 overload the network. 109 5. Assumptions Made and Limitations of such measurements 111 o The interactions of convergence and forwarding; testing is res- 112 tricted to events occurring within the control plane. Forwarding 113 performance is the primary focus in [INTERCONNECT] and it is 114 expected to be dealt with in work that ensues from [FIB-TERM]. 116 o Duplicate LSAs are Acknowledged Immediately. A few tests rely on 117 the property that duplicate LSA Acknowledgements are not delayed 118 but are done immediately. However if some implementation does not 119 acknowledge duplicate LSAs immediately on receipt, the testing 120 methods presented in [BENCHMARK] could give inaccurate measure- 121 ments. 123 o It is assumed that SPF is non-preemptive. If SPF is implemented so 124 that it can (and will be) preempted, the SPF measurements taken in 125 [BENCHMARK] would include the times that the SPF process is not 126 running ([BENCHMARK] measures the total time taken for SPF to run, 127 not the amount of time that SPF actually spends on the device's 128 processor), thus giving inaccurate measurements. 130 o Some implementations may be multithreaded or use a 131 multiprocess/multirouter model of OSPF. If because of this any of 132 the assumptions taken in measurement are violated in such a model, 133 it could lead to inaccurate measurements. 135 o The measurements resulting from the tests in [BENCHMARK] may not 136 provide the information required to deploy a device in a large 137 scale network. The tests described focus on individual components 138 of an OSPF implementation's performance, and it may be difficult 139 to combine the measurements in a way which accurately depicts a 140 device's performance in a large scale network. Further research is 141 required in this area. 143 o The measurements described in [BENCHMARK] should be used with 144 great care when comparing two different implementations of OSPF 145 from two different vendors. For instance, there are many other 146 factors than convergence speed which must be taken into considera- 147 tion when comparing different vendor's products, and it's diffi- 148 cult to align the resources available on one device to the 149 resources available on another device. 151 6. Observations on the Tests Described in [BENCHMARK] 153 Some observations taken while implementing the tests described in 154 [BENCHMARK] are noted in this section. 156 6.1. Measuring the SPF Processing Time Externally 158 The most difficult test to perform is the external measurement of the 159 time required to perform an SPF calculation, since the amount of time 160 between the first LSA which indicates a topology change and the 161 duplicate LSA is critical. If the duplicate LSA is sent too quickly, 162 it may be received before the device under test actually begins run- 163 ning SPF on the network change information. If the delay between the 164 two LSAs is too long, the device under test may finish SPF processing 165 before receiving the duplicate LSA. It is important to closely inves- 166 tigate any delays between the receipt of an LSA and the beginning of 167 an SPF calculation in the device under test; multiple tests with 168 various delays might be required to determine what delay needs to be 169 used to accurately measure the SPF calculation time. 171 Some implementations may force two intervals, the SPF hold time and 172 the SPF delay, between successive SPF calculations. If an SPF hold 173 time exists, it should be subtracted from the total SPF execution 174 time. If an SPF delay exists, it should be noted in the test results. 176 6.2. Noise in the Measurement Device 178 The device on which measurements are taken (not the device under 179 test) also adds noise to the test results, primarily in the form of 180 delay in packet processing and measurement output. The largest source 181 of noise is generally the delay between the receipt of packets by the 182 measuring device and the information about the packet reaching the 183 device's output, where the event can be measured. The following steps 184 may be taken to reduce this sampling noise: 186 o Increasing the number of samples taken will generally improve 187 the tester's ability to determine what is noise, and remove it 188 from the results. 190 o Try to take time-stamp for a packet as early as possible. 191 Depending on the operating system being used on the box, one 192 can instrument the kernel to take the time-stamp when the 193 interrupt is processed. This does not eliminate the noise com- 194 pletely, but at least reduces it. 196 o Keep the measurement box as lightly loaded as possible. 198 o Having an estimate of noise can also be useful. 200 The DUT also adds noise to the measurement. Points (a) and (c) 201 apply to the DUT as well. 203 6.3. Gaining an Understanding of the Implementation Improves Measure- 204 ments 206 While the tester will (generally) not have access to internal infor- 207 mation about the OSPF implementation being tested using [BENCHMARK], 208 the more thorough the tester's knowledge of the implementation is, 209 the more accurate the results of the tests will be. For instance, in 210 some implementations, the installation of routes in local routing 211 tables may occur while the SPF is being calculated, dramatically 212 impacting the time required to calculate the SPF. 214 6.4. Gaining an Understanding of the Tests Improves Measurements 216 One method which can be used to become familiar with the tests 217 described in [BENCHMARK] is to perform the tests on an OSPF implemen- 218 tation for which all the internal details are available, such as 219 GateD. While there is no assurance that any two implementations will 220 be similar, this will provide a better understanding of the tests 221 themselves. 223 7. LSA and Destination mix 225 In many OSPF benchmark tests, a generator injecting a number of LSAs 226 is called for. There are several areas in which injected LSAs can be 227 varied in testing: 229 o The number of destinations represented by the injected LSAs 231 Each destination represents a single reachable IP network; 232 these will be leaf nodes on the shortest path tree. The pri- 233 mary impact to performance should be the time required to 234 insert destinations in the local routing table and handling 235 the memory required to store the data. 237 o The types of LSAs injected 239 There are several types of LSAs which would be acceptable 240 under different situations; within an area, for instance, 241 type 1, 2, 3, 4, and 5 are likely to be received by a router. 242 Within a not-so-stubby area, however, type 7 LSAs would 243 replace the type 5 LSAs received. These sorts of characteri- 244 zations are important to note in any test results. 246 o The Number of LSAs injected 248 Within any injected set of information, the number of each 249 type of LSA injected is also important. This will impact the 250 shortest path algorithms ability to handle large numbers of 251 nodes, large shortest path first trees, etc. 253 o The Order of LSA Injection 255 The order in which LSAs are injected should not favor any 256 given data structure used for storing the LSA database on the 257 device under test. For instance, AS-External LSA's have AS 258 wide flooding scope; any Type-5 LSA originated is immediately 259 flooded to all neighbors. However the Type-4 LSA which 260 announces the ASBR as a border router is originated in an 261 area at SPF time (by ABR's on the edge of the area in which 262 the ASBR is). If SPF isn't scheduled immediately on the ABRs 263 originating the type 4 LSA, the type-4 LSA is sent after the 264 type-5 LSA's reach a router in the adjacent area. So routes 265 to the external destinations aren't immediately added to the 266 routers in the other areas. When the routers which already 267 have the type 5's receive the type-4 LSA, all the external 268 routes are added to the tree at the same time. This timing 269 could produce different results than a router receiving a 270 type 4 indicating the presence of a border router, followed 271 by the type 5's originated by that border router. 273 The ordering can be changed in various tests to provide 274 insight on the efficiency of storage within the DUT. Any such 275 changes in ordering should be noted in test results. 277 8. Tree Shape and the SPF Algorithm 279 The complexity of Dijkstra's algo depends on the data structure used 280 for storing vertices with their current minimum distances from the 281 source. The simplest structure is a list of vertices currently reach- 282 able from the source. Finding the minimum cost vertex then would take 283 O(size of the list). There will be O(n) such operations if we assume 284 that all the vertices are ultimately reachable from the source. More- 285 over, after the vertex with min cost is found, the algo iterates thru 286 all the edges of the vertex and updates cost of other vertices. With 287 an adjacency list representation, this step when iterated over all 288 the vertices, would take O(E) time. Thus, overall running time is: 290 O(sum(i:1, n)(size(list at level i) + E). 292 So, everything boils down to the size(list at level i). 294 If the graph is linear: 296 root 297 | 298 1 299 | 300 2 301 | 302 3 303 | 304 4 305 | 306 5 307 | 308 6 310 and source is a vertex on the end, then size(list at level i) 311 = 1 for all i. Moreover, E = n - 1. Therefore, running time 312 is O(n). 314 If graph is a balanced binary tree: 316 root 317 / \ 318 1 2 319 / \ / \ 320 3 4 5 6 322 size(list at level i) is a little complicated. First it 323 increases by 1 at each level upto a certain number, and then 324 goes down by 1. If we assumed that tree is a complete tree 325 (like the one in the draft) with k levels (1 to k), then 326 size(list) goes on like this: 1, 2, 3, 328 Then the number of edges E is still n - 1. It then turns out 329 that the run-time is O(n^2) for such a tree. 331 If graph is a complete graph (fully-connected mesh), then 332 size(list at level i) = n - i. Number of edges E = O(n^2). 333 Therefore, run-time is O(n^2). 335 shortest path first algorithm to compute the best paths 336 through the network need to be aware that the construction of 337 the tree may impact the performance of the algorithm. Best 338 practice would be to try and make any emulated network look 339 as much like a real network as possible, especially in the 340 area of the tree depth, the meshiness of the network, the 341 number of stub links verses transit links, and the number of 342 connections and nodes to process at each level within the 343 original tree. 345 9. Topology Generation 347 As the size of networks grows, it becomes more and more difficult to 348 actually create a large scale network on which to test the properties 349 of routing protocols and their implementations. In general, network 350 emulators are used to provide emulated topologies which can be adver- 351 tised to a device with varying conditions. Route generators either 352 tend to be a specialized device, a piece of software which runs on a 353 router, or a process that runs on another operating system, such as 354 Linux or another variant of Unix. 356 Some of the characteristics of this device should be: 358 o The ability to connect to the several devices using both point- 359 to-point and broadcast high speed media. Point-to-point links can 360 be emulated with high speed Ethernet as long as there is no hub or 361 other device in between the DUT and the route generator, and the 362 link is configured as a point-to-point link within OSPF. 364 o The ability to create a set of LSAs which appear to be a logical, 365 realistic topology. For instance, the generator should be able to 366 mix the number of point-to-point and broadcast links within the 367 emulated topology, and should be able to inject varying numbers of 368 externally reachable destinations. 370 o The ability to withdraw and add routing information into and from 371 the emulated topology to emulate links flapping. 373 o The ability to randomly order the LSAs representing the emulated 374 topology as they are advertised. 376 o The ability to log or otherwise measure the time between packets 377 transmitted and received. 379 o The ability to change the rate at which OSPF LSAs are transmitted. 381 o The generator and the collector should be fast enough so that they 382 are not bottle necks. The devices should also have a degree of 383 granularity of measurement atleast as small as desired from the 384 test results. 386 10. Acknowledgements 388 Thanks to Howard Berkowitz, (hcb@clark.net) and the rest of the BGP 389 benchmarking team for their support and to Kevin 390 Dubray(kdubray@juniper.net) who realized the need of this draft. 392 11. Normative References 394 [BENCHMARK] 395 Manral, V., "Benchmarking Basic OSPF Single Router Control Plane 396 Convergence", draft-bmwg-ospfconv-intraarea-05, March 2003 398 [TERM]Manral, V., "OSPF Convergence Testing Terminiology and Concepts", 399 draft-bmwg-ospfconv-term-04, March 2003 401 12. Informative References 403 [INTERCONNECT] 404 Bradner, S., McQuaid, J., "Benchmarking Methodology for Network 405 Interconnect Devices", RFC2544, March 1999. 407 [FIB-TERM] 408 Trotter, G., "Terminology for Forwarding Information Base (FIB) 409 based Router Performance", RFC3222, October 2001. 411 13. Authors' Addresses 412 Vishwas Manral 413 Netplane Systems 414 189 Prashasan Nagar 415 Road number 72 416 Jubilee Hills 417 Hyderabad, India 419 vmanral@netplane.com 421 Russ White 422 Cisco Systems, Inc. 423 7025 Kit Creek Rd. 424 Research Triangle Park, NC 27709 426 riw@cisco.com 428 Aman Shaikh 429 University of California 430 School of Engineering 431 1156 High Street 432 Santa Cruz, CA 95064 433 aman@soe.ucsc.edu