ceph/src/test/erasure-code/ceph_erasure_code_benchmark.cc

   1 // -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*-
   2 // vim: ts=8 sw=2 smarttab
   3 /*
   4  * Ceph distributed storage system
   5  *
   6  * Copyright (C) 2013,2014 Cloudwatt <libre.licensing@cloudwatt.com>
   7  * Copyright (C) 2014 Red Hat <contact@redhat.com>
   8  *
   9  * Author: Loic Dachary <loic@dachary.org>
  10  *
  11  *  This library is free software; you can redistribute it and/or
  12  *  modify it under the terms of the GNU Lesser General Public
  13  *  License as published by the Free Software Foundation; either
  14  *  version 2.1 of the License, or (at your option) any later version.
  15  *
  16  */
  17
  18 #include <boost/scoped_ptr.hpp>
  19 #include <boost/lexical_cast.hpp>
  20 #include <boost/program_options/option.hpp>
  21 #include <boost/program_options/options_description.hpp>
  22 #include <boost/program_options/variables_map.hpp>
  23 #include <boost/program_options/cmdline.hpp>
  24 #include <boost/program_options/parsers.hpp>
  25 #include <boost/algorithm/string.hpp>
  26
  27 #include "global/global_context.h"
  28 #include "global/global_init.h"
  29 #include "common/ceph_argparse.h"
  30 #include "common/ceph_context.h"
  31 #include "common/config.h"
  32 #include "common/Clock.h"
  33 #include "include/utime.h"
  34 #include "erasure-code/ErasureCodePlugin.h"
  35 #include "erasure-code/ErasureCode.h"
  36 #include "ceph_erasure_code_benchmark.h"
  37
  38 namespace po = boost::program_options;
  39
  40 int ErasureCodeBench::setup(int argc, char** argv) {
  41
  42   po::options_description desc("Allowed options");
  43   desc.add_options()
  44     ("help,h", "produce help message")
  45     ("verbose,v", "explain what happens")
  46     ("size,s", po::value<int>()->default_value(1024 * 1024),
  47      "size of the buffer to be encoded")
  48     ("iterations,i", po::value<int>()->default_value(1),
  49      "number of encode/decode runs")
  50     ("plugin,p", po::value<string>()->default_value("jerasure"),
  51      "erasure code plugin name")
  52     ("workload,w", po::value<string>()->default_value("encode"),
  53      "run either encode or decode")
  54     ("erasures,e", po::value<int>()->default_value(1),
  55      "number of erasures when decoding")
  56     ("erased", po::value<vector<int> >(),
  57      "erased chunk (repeat if more than one chunk is erased)")
  58     ("erasures-generation,E", po::value<string>()->default_value("random"),
  59      "If set to 'random', pick the number of chunks to recover (as specified by "
  60      " --erasures) at random. If set to 'exhaustive' try all combinations of erasures "
  61      " (i.e. k=4,m=3 with one erasure will try to recover from the erasure of "
  62      " the first chunk, then the second etc.)")
  63     ("parameter,P", po::value<vector<string> >(),
  64      "add a parameter to the erasure code profile")
  65     ;
  66
  67   po::variables_map vm;
  68   po::parsed_options parsed =
  69     po::command_line_parser(argc, argv).options(desc).allow_unregistered().run();
  70   po::store(
  71     parsed,
  72     vm);
  73   po::notify(vm);
  74
  75   vector<const char *> ceph_options;
  76   vector<string> ceph_option_strings = po::collect_unrecognized(
  77     parsed.options, po::include_positional);
  78   ceph_options.reserve(ceph_option_strings.size());
  79   for (vector<string>::iterator i = ceph_option_strings.begin();
  80        i != ceph_option_strings.end();
  81        ++i) {
  82     ceph_options.push_back(i->c_str());
  83   }
  84
  85   cct = global_init(
  86     NULL, ceph_options, CEPH_ENTITY_TYPE_CLIENT,
  87     CODE_ENVIRONMENT_UTILITY,
  88     CINIT_FLAG_NO_DEFAULT_CONFIG_FILE);
  89   common_init_finish(g_ceph_context);
  90   g_ceph_context->_conf.apply_changes(nullptr);
  91
  92   if (vm.count("help")) {
  93     cout << desc << std::endl;
  94     return 1;
  95   }
  96
  97   if (vm.count("parameter")) {
  98     const vector<string> &p = vm["parameter"].as< vector<string> >();
  99     for (vector<string>::const_iterator i = p.begin();
 100          i != p.end();
 101          ++i) {
 102       std::vector<std::string> strs;
 103       boost::split(strs, *i, boost::is_any_of("="));
 104       if (strs.size() != 2) {
 105         cerr << "--parameter " << *i << " ignored because it does not contain exactly one =" << endl;
 106       } else {
 107         profile[strs[0]] = strs[1];
 108       }
 109     }
 110   }
 111
 112   in_size = vm["size"].as<int>();
 113   max_iterations = vm["iterations"].as<int>();
 114   plugin = vm["plugin"].as<string>();
 115   workload = vm["workload"].as<string>();
 116   erasures = vm["erasures"].as<int>();
 117   if (vm.count("erasures-generation") > 0 &&
 118       vm["erasures-generation"].as<string>() == "exhaustive")
 119     exhaustive_erasures = true;
 120   else
 121     exhaustive_erasures = false;
 122   if (vm.count("erased") > 0)
 123     erased = vm["erased"].as<vector<int> >();
 124
 125   k = stoi(profile["k"]);
 126   m = stoi(profile["m"]);
 127
 128   if (k <= 0) {
 129     cout << "parameter k is " << k << ". But k needs to be > 0." << endl;
 130     return -EINVAL;
 131   } else if ( m < 0 ) {
 132     cout << "parameter m is " << m << ". But m needs to be >= 0." << endl;
 133     return -EINVAL;
 134   }
 135
 136   verbose = vm.count("verbose") > 0 ? true : false;
 137
 138   return 0;
 139 }
 140
 141 int ErasureCodeBench::run() {
 142   ErasureCodePluginRegistry &instance = ErasureCodePluginRegistry::instance();
 143   instance.disable_dlclose = true;
 144
 145   if (workload == "encode")
 146     return encode();
 147   else
 148     return decode();
 149 }
 150
 151 int ErasureCodeBench::encode()
 152 {
 153   ErasureCodePluginRegistry &instance = ErasureCodePluginRegistry::instance();
 154   ErasureCodeInterfaceRef erasure_code;
 155   stringstream messages;
 156   int code = instance.factory(plugin,
 157                               g_conf().get_val<std::string>("erasure_code_dir"),
 158                               profile, &erasure_code, &messages);
 159   if (code) {
 160     cerr << messages.str() << endl;
 161     return code;
 162   }
 163
 164   bufferlist in;
 165   in.append(string(in_size, 'X'));
 166   in.rebuild_aligned(ErasureCode::SIMD_ALIGN);
 167   set<int> want_to_encode;
 168   for (int i = 0; i < k + m; i++) {
 169     want_to_encode.insert(i);
 170   }
 171   utime_t begin_time = ceph_clock_now();
 172   for (int i = 0; i < max_iterations; i++) {
 173     map<int,bufferlist> encoded;
 174     code = erasure_code->encode(want_to_encode, in, &encoded);
 175     if (code)
 176       return code;
 177   }
 178   utime_t end_time = ceph_clock_now();
 179   cout << (end_time - begin_time) << "\t" << (max_iterations * (in_size / 1024)) << endl;
 180   return 0;
 181 }
 182
 183 static void display_chunks(const map<int,bufferlist> &chunks,
 184                            unsigned int chunk_count) {
 185   cout << "chunks ";
 186   for (unsigned int chunk = 0; chunk < chunk_count; chunk++) {
 187     if (chunks.count(chunk) == 0) {
 188       cout << "(" << chunk << ")";
 189     } else {
 190       cout << " " << chunk << " ";
 191     }
 192     cout << " ";
 193   }
 194   cout << "(X) is an erased chunk" << endl;
 195 }
 196
 197 int ErasureCodeBench::decode_erasures(const map<int,bufferlist> &all_chunks,
 198                                       const map<int,bufferlist> &chunks,
 199                                       unsigned i,
 200                                       unsigned want_erasures,
 201                                       ErasureCodeInterfaceRef erasure_code)
 202 {
 203   int code = 0;
 204
 205   if (want_erasures == 0) {
 206     if (verbose)
 207       display_chunks(chunks, erasure_code->get_chunk_count());
 208     set<int> want_to_read;
 209     for (unsigned int chunk = 0; chunk < erasure_code->get_chunk_count(); chunk++)
 210       if (chunks.count(chunk) == 0)
 211         want_to_read.insert(chunk);
 212
 213     map<int,bufferlist> decoded;
 214     code = erasure_code->decode(want_to_read, chunks, &decoded, 0);
 215     if (code)
 216       return code;
 217     for (set<int>::iterator chunk = want_to_read.begin();
 218          chunk != want_to_read.end();
 219          ++chunk) {
 220       if (all_chunks.find(*chunk)->second.length() != decoded[*chunk].length()) {
 221         cerr << "chunk " << *chunk << " length=" << all_chunks.find(*chunk)->second.length()
 222              << " decoded with length=" << decoded[*chunk].length() << endl;
 223         return -1;
 224       }
 225       bufferlist tmp = all_chunks.find(*chunk)->second;
 226       if (!tmp.contents_equal(decoded[*chunk])) {
 227         cerr << "chunk " << *chunk
 228              << " content and recovered content are different" << endl;
 229         return -1;
 230       }
 231     }
 232     return 0;
 233   }
 234
 235   for (; i < erasure_code->get_chunk_count(); i++) {
 236     map<int,bufferlist> one_less = chunks;
 237     one_less.erase(i);
 238     code = decode_erasures(all_chunks, one_less, i + 1, want_erasures - 1, erasure_code);
 239     if (code)
 240       return code;
 241   }
 242
 243   return 0;
 244 }
 245
 246 int ErasureCodeBench::decode()
 247 {
 248   ErasureCodePluginRegistry &instance = ErasureCodePluginRegistry::instance();
 249   ErasureCodeInterfaceRef erasure_code;
 250   stringstream messages;
 251   int code = instance.factory(plugin,
 252                               g_conf().get_val<std::string>("erasure_code_dir"),
 253                               profile, &erasure_code, &messages);
 254   if (code) {
 255     cerr << messages.str() << endl;
 256     return code;
 257   }
 258
 259   bufferlist in;
 260   in.append(string(in_size, 'X'));
 261   in.rebuild_aligned(ErasureCode::SIMD_ALIGN);
 262
 263   set<int> want_to_encode;
 264   for (int i = 0; i < k + m; i++) {
 265     want_to_encode.insert(i);
 266   }
 267
 268   map<int,bufferlist> encoded;
 269   code = erasure_code->encode(want_to_encode, in, &encoded);
 270   if (code)
 271     return code;
 272
 273   set<int> want_to_read = want_to_encode;
 274
 275   if (erased.size() > 0) {
 276     for (vector<int>::const_iterator i = erased.begin();
 277          i != erased.end();
 278          ++i)
 279       encoded.erase(*i);
 280     display_chunks(encoded, erasure_code->get_chunk_count());
 281   }
 282
 283   utime_t begin_time = ceph_clock_now();
 284   for (int i = 0; i < max_iterations; i++) {
 285     if (exhaustive_erasures) {
 286       code = decode_erasures(encoded, encoded, 0, erasures, erasure_code);
 287       if (code)
 288         return code;
 289     } else if (erased.size() > 0) {
 290       map<int,bufferlist> decoded;
 291       code = erasure_code->decode(want_to_read, encoded, &decoded, 0);
 292       if (code)
 293         return code;
 294     } else {
 295       map<int,bufferlist> chunks = encoded;
 296       for (int j = 0; j < erasures; j++) {
 297         int erasure;
 298         do {
 299           erasure = rand() % ( k + m );
 300         } while(chunks.count(erasure) == 0);
 301         chunks.erase(erasure);
 302       }
 303       map<int,bufferlist> decoded;
 304       code = erasure_code->decode(want_to_read, chunks, &decoded, 0);
 305       if (code)
 306         return code;
 307     }
 308   }
 309   utime_t end_time = ceph_clock_now();
 310   cout << (end_time - begin_time) << "\t" << (max_iterations * (in_size / 1024)) << endl;
 311   return 0;
 312 }
 313
 314 int main(int argc, char** argv) {
 315   ErasureCodeBench ecbench;
 316   try {
 317     int err = ecbench.setup(argc, argv);
 318     if (err)
 319       return err;
 320     return ecbench.run();
 321   } catch(po::error &e) {
 322     cerr << e.what() << endl;
 323     return 1;
 324   }
 325 }
 326
 327 /*
 328  * Local Variables:
 329  * compile-command: "cd ../../../build ; make -j4 ceph_erasure_code_benchmark &&
 330  *   valgrind --tool=memcheck --leak-check=full \
 331  *      ./bin/ceph_erasure_code_benchmark \
 332  *      --plugin jerasure \
 333  *      --parameter directory=lib \
 334  *      --parameter technique=reed_sol_van \
 335  *      --parameter k=2 \
 336  *      --parameter m=2 \
 337  *      --iterations 1
 338  * "
 339  * End:
 340  */