ftsbench.c

   1 /*
   2  * Copyright (c) 2006 Teodor Sigaev <teodor@sigaev.ru>
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *        notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *        notice, this list of conditions and the following disclaimer in the
  12  *        documentation and/or other materials provided with the distribution.
  13  * 3. Neither the name of the author nor the names of any co-contributors
  14  *        may be used to endorse or promote products derived from this software
  15  *        without specific prior written permission.
  16  *
  17  * THIS SOFTWARE IS PROVIDED BY CONTRIBUTORS ``AS IS'' AND ANY EXPRESS
  18  * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
  19  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  20  * ARE DISCLAIMED. IN NO EVENT SHALL CONTRIBUTORS BE LIABLE FOR ANY
  21  * DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE
  23  * GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
  24  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER
  25  * IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
  26  * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN
  27  * IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
  28  */
  29
  30 #include <stdio.h>
  31 #include <stdlib.h>
  32 #include <unistd.h>
  33 #include <string.h>
  34 #include <errno.h>
  35 #include <sys/time.h>
  36
  37 #include "ftsbench.h"
  38
  39 typedef enum RDBMS {
  40         PostgreSQL = 0,
  41         MySQL = 1,
  42         NULLSQL
  43 } RDBMS;
  44
  45 typedef struct RDBMSDesc {
  46         RDBMS   rdbms;
  47         char    *shortname;
  48         char    *longname;
  49         ftsDB*  (*init)(char *);
  50 } RDBMSDesc;
  51
  52 static RDBMSDesc DBDesc[] = {
  53         { PostgreSQL, "pgsql", "PostgreSQL", PGInit },
  54         { MySQL,          "mysql", "MySQL",      MYInit },
  55         { NULLSQL,        NULL,    NULL,         NULL   }
  56 };
  57
  58 static void
  59 usage() {
  60         char buf[1024];
  61         int i, first=0;
  62
  63         *buf = '\0';
  64         for(i=0; DBDesc[i].rdbms != NULLSQL; i++) {
  65                 if ( DBDesc[i].init == NULL )
  66                         continue;
  67                 if ( first != 0 )
  68                         strcat(buf, ", ");
  69                 strcat(buf, DBDesc[i].shortname);
  70                 if ( first == 0 )
  71                         strcat(buf, "(default)");
  72                 first++;
  73         }
  74
  75         fputs(
  76                 "ftsbench - full text search benchmark ofr RDBMS\n"
  77                 "Initialization of DB:\n"
  78                 "\tftsbench -i [-b RDBMS] [-n NUMROW] [-l LEXFILE] [-g GAMMAFILE] [-f FLAGS] [-q] -d DBNAME\n"
  79                 "FLAGS are comma-separate list of:\n"
  80                 "       gin  - use GIN index\n"
  81                 "       gist - use GiST index\n"
  82                 "       func - use functional index\n",
  83                 stdout
  84         );
  85         fputs(
  86                 "Run tests:\n"
  87                 "\tftsbench [-b RDBMS] [-c NCLIENTS] [-n NUMQUERY] [-l LEXFILE] [-g GAMMAFILE] [-f FLAGS] [-q] -d DBNAME\n"
  88                 "FLAGS are comma-separate list of:\n"
  89                 "       and  - AND'ing lexemes in query (default)\n"
  90                 "       or   - OR'ing lexemes in query\n"
  91                 "       sort - sort result of query\n"
  92                 "Options are:\n"
  93                 "       -b RDBMS\t- type of DB: ",
  94                 stdout
  95         );
  96         fputs( buf, stdout );
  97         fputs(
  98                 "\n"
  99                 "       -l LEXFILE\t- file with words and its frequents\n"
 100                 "       -g GAMMAFILE\t- file with doc's length distribution\n",
 101                 stdout
 102         );
 103         exit(1);
 104 }
 105
 106 static RDBMS
 107 getRDBMS(char *name) {
 108         int     i;
 109
 110         for(i=0; DBDesc[i].rdbms != NULLSQL; i++) {
 111                 if ( name == NULL ) {
 112                         if ( DBDesc[i].init )
 113                                 return DBDesc[i].rdbms;
 114                 } else if ( strcasecmp(name,DBDesc[i].shortname) == 0 ) {
 115                         if ( DBDesc[i].init == NULL ) {
 116                                 fprintf(stderr,"Support of '%s' isn't compiled-in\n", DBDesc[i].longname);
 117                                 exit(1);
 118                         }
 119                         return DBDesc[i].rdbms;
 120                 }
 121         }
 122
 123         fprintf(stderr,"Can't find a RDBMS\n");
 124         exit(1);
 125
 126         return NULLSQL;
 127 }
 128
 129 static int
 130 getFLAGS(char *flg) {
 131         int flags = 0;
 132
 133         if ( strcasestr(flg,"gist") )
 134                 flags |= FLG_GIST;
 135         if ( strcasestr(flg,"gin") )
 136                 flags |= FLG_GIN;
 137         if ( strcasestr(flg,"func") )
 138                 flags |= FLG_FUNC;
 139         if ( strcasestr(flg,"and") )
 140                 flags |= FLG_AND;
 141         if ( strcasestr(flg,"or") )
 142                 flags |= FLG_OR;
 143         if ( strcasestr(flg,"sort") )
 144                 flags |= FLG_SORT;
 145
 146         if ( (flags & FLG_GIST) && (flags & FLG_GIN) ) {
 147                 fprintf(stderr,"GIN and GiST flags are mutually exclusive\n");
 148                 exit(1);
 149         }
 150         if ( (flags & FLG_AND) && (flags & FLG_OR) ) {
 151                 fprintf(stderr,"AND and OR flags are mutually exclusive\n");
 152                 exit(1);
 153         }
 154
 155         return flags;
 156 }
 157
 158 static ftsDB **
 159 initConnections(RDBMS rdbms, int n, char *connstr) {
 160         ftsDB   **dbs = (ftsDB**)malloc(sizeof(ftsDB*) * n);
 161         int i;
 162
 163         if (!dbs) {
 164                 fprintf(stderr,"Not enough mwmory\n");
 165                 exit(1);
 166         }
 167
 168         for(i=0;i<n;i++) {
 169                 dbs[i] = DBDesc[rdbms].init(connstr);
 170                 pthread_mutex_init(&dbs[i]->nqueryMutex, NULL);
 171         }
 172
 173         return dbs;
 174 }
 175
 176 static double
 177 timediff(struct timeval *begin, struct timeval *end) {
 178     return ((double)( end->tv_sec - begin->tv_sec )) + ( (double)( end->tv_usec-begin->tv_usec ) ) / 1.0e+6;
 179 }
 180
 181 static double
 182 elapsedtime(struct timeval *begin) {
 183     struct timeval end;
 184         gettimeofday(&end,NULL);
 185         return timediff(begin,&end);
 186 }
 187
 188 static int benchFlags  = 0;
 189 static int benchCount  = 0;
 190 static pthread_cond_t condFinish = PTHREAD_COND_INITIALIZER;
 191 static pthread_mutex_t mutexFinish = PTHREAD_MUTEX_INITIALIZER;
 192 static pthread_mutex_t mutexWordGen = PTHREAD_MUTEX_INITIALIZER;
 193
 194 static void*
 195 execBench(void *in) {
 196         ftsDB *db = (ftsDB*)in;
 197         int i;
 198         char **words;
 199
 200         for(i=0;i<benchCount;i++) {
 201                 /*
 202                  * generate_querywords() isn't a thread safe
 203                  */
 204                 pthread_mutex_lock( &mutexWordGen );
 205                 words = generate_querywords();
 206                 pthread_mutex_unlock( &mutexWordGen );
 207
 208                 db->execQuery(db, words, benchFlags);
 209                 free(words);
 210         }
 211
 212         /*
 213          * send message about exitting
 214          */
 215     pthread_mutex_lock( &mutexFinish );
 216         pthread_cond_broadcast( &condFinish );
 217         pthread_mutex_unlock( &mutexFinish );
 218
 219         return NULL;
 220 }
 221
 222 extern char *optarg;
 223
 224 int
 225 main(int argn, char *argv[]) {
 226         int             initMode = 0;
 227         int             n = 0, nclients = 1;
 228         char    *lex = NULL;
 229         char    *doc = NULL;
 230         char    *dbname = NULL;
 231         RDBMS   rdbms = NULLSQL;
 232         int             flags = 0;
 233         int i;
 234         int             quiet = 0;
 235         StringBuf       b = {NULL,0,0};
 236
 237         while((i=getopt(argn,argv,"ib:n:l:g:d:c:hf:q")) != EOF) {
 238                 switch(i) {
 239                         case 'i': initMode = 1; break;
 240                         case 'b': rdbms = getRDBMS(optarg); break;
 241                         case 'n': n=atoi(optarg); break;
 242                         case 'c': nclients=atoi(optarg); break;
 243                         case 'l': lex = strdup(optarg); break;
 244                         case 'g': doc = strdup(optarg); break;
 245                         case 'd': dbname = strdup(optarg); break;
 246                         case 'f': flags = getFLAGS(optarg); break;
 247                         case 'q': quiet = 1; break;
 248                         case 'h':
 249                         default:
 250                                 usage();
 251                 }
 252         }
 253
 254         if (rdbms == NULLSQL)
 255                 rdbms = getRDBMS(NULL);
 256
 257         if ( dbname == NULL || n<0 || nclients<1 )
 258                 usage();
 259
 260         printf("Running with '%s' RDBMS\n", DBDesc[ rdbms ].longname);
 261
 262         if ( initMode ) {
 263                 ftsDB   *db = *initConnections(rdbms, 1, dbname);
 264                 time_t  prev;
 265
 266                 if (!lex)  lex = "gendata/lex";
 267                 if (!doc)  doc = "gendata/gamma-lens";
 268                 finnegan_init(lex, doc);
 269
 270                 db->startCreateScheme(db, flags);
 271                 prev = time(NULL);
 272                 for(i=0;i<n;i++) {
 273                         generate_doc(&b);
 274                         db->InsertRow(db, i+1, b.str);
 275                         if ( !quiet && prev!=time(NULL) ) {
 276                                 printf("\r%d(%.02f%%) rows inserted", i, (100.0*i)/n);
 277                                 fflush(stdout);
 278                                 prev = time(NULL);
 279                         }
 280                 }
 281                 printf("%s%d(100.00%%) rows inserted. Finalyze insertion... ",
 282                         (quiet) ? "" : "\r", i);
 283                 fflush(stdout);
 284                 db->finishCreateScheme(db);
 285                 printf("done\n");
 286                 db->Close(db);
 287         } else {
 288                 ftsDB   **dbs = initConnections(rdbms, nclients, dbname);
 289                 pthread_t       *tid = (pthread_t*)malloc( sizeof(pthread_t) * nclients);
 290                 struct  timeval begin;
 291                 double  elapsed;
 292                 int     total=0;
 293                 struct      timespec  sleepTo = { 0, 0 };
 294
 295                 /*
 296                  * startup generator
 297                  */
 298                 if (!lex)  lex = "gendata/query-lex";
 299                 if (!doc)  doc = "gendata/query-lens";
 300                 finnegan_init(lex, doc);
 301
 302                 /*
 303                  * Initial query
 304                  */
 305                 if ( !quiet ) {
 306                         printf("\r0(0.00%%) queries proceed");
 307                         fflush(stdout);
 308                 }
 309                 benchFlags = flags;
 310                 benchCount = n;
 311
 312                 gettimeofday(&begin,NULL);
 313
 314         pthread_mutex_lock( &mutexFinish );
 315                 for(i=0;i<nclients;i++) {
 316                         if ( pthread_create(tid+i, NULL, execBench, (void*)dbs[i]) != 0 ) {
 317                                 fprintf(stderr,"pthread_create failed: %s\n", strerror(errno));
 318                                 exit(1);
 319                         }
 320                 }
 321
 322                 for(;;) {
 323                         int res, ntogo = 0;
 324
 325                         total = 0;
 326                         for(i=0;i<nclients;i++) {
 327                                 pthread_mutex_lock(&dbs[i]->nqueryMutex);
 328                                 total +=dbs[i]->nquery;
 329                                 if ( dbs[i]->nquery < n )
 330                                         ntogo++;
 331                                 pthread_mutex_unlock(&dbs[i]->nqueryMutex);
 332                         }
 333
 334                         if ( ntogo == 0 )
 335                                 break;
 336
 337                         if ( !quiet ) {
 338                                 printf("\r%d(%.02f%%) queries proceed", total, (100.0*(float)total)/(nclients * n));
 339                                 fflush(stdout);
 340                         }
 341
 342                         sleepTo.tv_sec = time(NULL) + 1;
 343                         res = pthread_cond_timedwait( &condFinish, &mutexFinish, &sleepTo );
 344
 345                         if ( !(res == ETIMEDOUT || res == 0) ) {
 346                                 fprintf(stderr,"pthread_cond_timedwait failed: %s\n", strerror(errno));
 347                                 exit(1);
 348                         }
 349                 }
 350                 elapsed = elapsedtime(&begin);
 351                 pthread_mutex_unlock( &mutexFinish );
 352
 353                 for(i=0;i<nclients;i++) {
 354                         pthread_join(tid[i], NULL);
 355                         dbs[i]->Close(dbs[i]);
 356                 }
 357
 358                 printf("%s%d(%.02f%%) queries proceed\n",
 359                         (quiet) ? "" : "\r", total, (100.0*(float)total)/(nclients * n));
 360                 printf("Total time: %.02f sec, Queries per second: %.02f\n", elapsed, total/elapsed);
 361                 fflush(stdout);
 362         }
 363
 364         return 0;
 365 }