1<?xml version="1.0" encoding="UTF-8" standalone="no"?> 2<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> 3<html xmlns="http://www.w3.org/1999/xhtml"> 4 <head> 5 <meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /> 6 <title>Access Methods</title> 7 <link rel="stylesheet" href="gettingStarted.css" type="text/css" /> 8 <meta name="generator" content="DocBook XSL Stylesheets V1.73.2" /> 9 <link rel="start" href="index.html" title="Getting Started with Berkeley DB" /> 10 <link rel="up" href="introduction.html" title="Chapter 1. Introduction to Berkeley DB" /> 11 <link rel="prev" href="concepts.html" title="Berkeley DB Concepts" /> 12 <link rel="next" href="databaseLimits.html" title="Database Limits and Portability" /> 13 </head> 14 <body> 15 <div class="navheader"> 16 <table width="100%" summary="Navigation header"> 17 <tr> 18 <th colspan="3" align="center">Access Methods</th> 19 </tr> 20 <tr> 21 <td width="20%" align="left"><a accesskey="p" href="concepts.html">Prev</a> </td> 22 <th width="60%" align="center">Chapter 1. Introduction to Berkeley DB </th> 23 <td width="20%" align="right"> <a accesskey="n" href="databaseLimits.html">Next</a></td> 24 </tr> 25 </table> 26 <hr /> 27 </div> 28 <div class="sect1" lang="en" xml:lang="en"> 29 <div class="titlepage"> 30 <div> 31 <div> 32 <h2 class="title" style="clear: both"><a id="accessmethods"></a>Access Methods</h2> 33 </div> 34 </div> 35 </div> 36 <div class="toc"> 37 <dl> 38 <dt> 39 <span class="sect2"> 40 <a href="accessmethods.html#selectAM">Selecting Access Methods</a> 41 </span> 42 </dt> 43 <dt> 44 <span class="sect2"> 45 <a href="accessmethods.html#BTreeVSHash">Choosing between BTree and Hash</a> 46 </span> 47 </dt> 48 <dt> 49 <span class="sect2"> 50 <a href="accessmethods.html#QueueVSRecno">Choosing between Queue and Recno</a> 51 </span> 52 </dt> 53 </dl> 54 </div> 55 <p> 56 While this manual will focus primarily on the BTree access method, it is 57 still useful to briefly describe all of the access methods that DB 58 makes available. 59 </p> 60 <p> 61 Note that an access method can be selected only when the database is 62 created. Once selected, actual API usage is generally 63 identical across all access methods. That is, while some 64 exceptions exist, mechanically you interact with the library in the same 65 way regardless of which access method you have selected. 66 </p> 67 <p> 68 The access method that you should choose is gated first by what you want 69 to use as a key, and then secondly by the performance that you see 70 for a given access method. 71 </p> 72 <p> 73 The following are the available access methods: 74 </p> 75 <div class="informaltable"> 76 <table border="1" width="80%"> 77 <colgroup> 78 <col align="left" /> 79 <col align="left" /> 80 </colgroup> 81 <thead> 82 <tr> 83 <th align="center">Access Method</th> 84 <th align="center">Description</th> 85 </tr> 86 </thead> 87 <tbody> 88 <tr> 89 <td align="left">BTree</td> 90 <td align="left" valign="top"> 91 <p> 92 Data is stored in a sorted, balanced tree structure. 93 Both the key and the data for BTree records can be 94 arbitrarily complex. That is, they can contain single values 95 such as an integer or a string, or complex types such as a 96 structure. Also, although not the default 97 behavior, it is possible for two records to 98 use keys that compare as equals. When this occurs, the 99 records are considered to be duplicates of one another. 100 </p> 101 </td> 102 </tr> 103 <tr> 104 <td align="left">Hash</td> 105 <td align="left" valign="top"> 106 <p> 107 Data is stored in an extended linear hash table. Like 108 BTree, the key and the data used for Hash records can be of 109 arbitrarily complex data. Also, like BTree, duplicate 110 records are optionally supported. 111 </p> 112 </td> 113 </tr> 114 <tr> 115 <td align="left">Queue</td> 116 <td align="left" valign="top"> 117 <p> 118 Data is stored in a queue as fixed-length records. Each 119 record uses a logical record number as its key. This access 120 method is designed for fast inserts at the tail of the 121 queue, and it has a special operation that deletes and 122 returns a record from the head of the queue. 123 </p> 124 <p> 125 This access method is unusual in that it provides record 126 level locking. This can provide 127 beneficial performance improvements in applications 128 requiring concurrent access to the queue. 129 </p> 130 </td> 131 </tr> 132 <tr> 133 <td align="left">Recno</td> 134 <td align="left" valign="top"> 135 <p> 136 Data is stored in either fixed or variable-length records. 137 Like Queue, Recno records use logical record numbers as keys. 138 </p> 139 </td> 140 </tr> 141 </tbody> 142 </table> 143 </div> 144 <div class="sect2" lang="en" xml:lang="en"> 145 <div class="titlepage"> 146 <div> 147 <div> 148 <h3 class="title"><a id="selectAM"></a>Selecting Access Methods</h3> 149 </div> 150 </div> 151 </div> 152 <p> 153 To select an access method, you should first consider what you want 154 to use as a key for you database records. If you want to use 155 arbitrary data (even strings), then you should use either BTree or 156 Hash. If you want to use logical record numbers (essentially 157 integers) then you should use Queue or Recno. 158 </p> 159 <p> 160 Once you have made this decision, you must choose between either 161 BTree or Hash, or Queue or Recno. This decision is described next. 162 </p> 163 </div> 164 <div class="sect2" lang="en" xml:lang="en"> 165 <div class="titlepage"> 166 <div> 167 <div> 168 <h3 class="title"><a id="BTreeVSHash"></a>Choosing between BTree and Hash</h3> 169 </div> 170 </div> 171 </div> 172 <p> 173 For small working datasets that fit entirely in memory, there is no 174 difference between BTree and Hash. Both will perform just as well 175 as the other. In this situation, you might just as well use BTree, 176 if for no other reason than the majority of DB applications use 177 BTree. 178 </p> 179 <p> 180 Note that the main concern here is your 181 working dataset, not your entire dataset. Many applications maintain 182 large amounts of information but only need to access some small 183 portion of that data with any frequency. So what you want to 184 consider is the data that you will routinely use, not the sum total 185 of all the data managed by your application. 186 </p> 187 <p> 188 However, as your working dataset grows to the point 189 where you cannot fit it all into memory, then you need to take more 190 care when choosing your access method. Specifically, choose: 191 </p> 192 <div class="itemizedlist"> 193 <ul type="disc"> 194 <li> 195 <p> 196 BTree if your keys have some locality of reference. That is, 197 if they sort well and you can expect that a query for a 198 given key will likely be followed by a query for one of its 199 neighbors. 200 </p> 201 </li> 202 <li> 203 <p> 204 Hash if your dataset is extremely large. For any given 205 access method, DB must maintain a certain amount of internal 206 information. However, the amount of information that DB 207 must maintain for BTree is much greater than for Hash. The 208 result is that as your dataset grows, this internal 209 information can dominate the cache to the point where there 210 is relatively little space left for application data. 211 As a result, BTree can be forced to perform disk I/O much more 212 frequently than would Hash given the same amount of data. 213 </p> 214 <p> 215 Moreover, if your dataset becomes so large that DB will 216 almost certainly have to perform disk I/O to satisfy a 217 random request, then Hash will definitely out perform BTree 218 because it has fewer internal records to search through than 219 does BTree. 220 </p> 221 </li> 222 </ul> 223 </div> 224 </div> 225 <div class="sect2" lang="en" xml:lang="en"> 226 <div class="titlepage"> 227 <div> 228 <div> 229 <h3 class="title"><a id="QueueVSRecno"></a>Choosing between Queue and Recno</h3> 230 </div> 231 </div> 232 </div> 233 <p> 234 Queue or Recno are used when the application wants to use logical 235 record numbers for the primary database key. Logical record numbers 236 are essentially integers that uniquely identify the database 237 record. They can be either mutable or fixed, where a mutable record 238 number is one that might change as database records are stored or 239 deleted. Fixed logical record numbers never change regardless of 240 what database operations are performed. 241 </p> 242 <p> 243 When deciding between Queue and Recno, choose: 244 </p> 245 <div class="itemizedlist"> 246 <ul type="disc"> 247 <li> 248 <p> 249 Queue if your application requires high degrees of 250 concurrency. Queue provides record-level locking (as opposed 251 to the page-level locking that the other access methods 252 use), and this can result in significantly faster throughput 253 for highly concurrent applications. 254 </p> 255 <p> 256 Note, however, that Queue provides support only for fixed 257 length records. So if the size of the data that you want to 258 store varies widely from record to record, you should 259 probably choose an access method other than Queue. 260 </p> 261 </li> 262 <li> 263 <p> 264 Recno if you want mutable record numbers. Queue is only 265 capable of providing fixed record numbers. Also, Recno 266 provides support for databases whose permanent storage is a 267 flat text file. This is useful for applications looking for 268 fast, temporary storage while the data is being read or 269 modified. 270 </p> 271 </li> 272 </ul> 273 </div> 274 </div> 275 </div> 276 <div class="navfooter"> 277 <hr /> 278 <table width="100%" summary="Navigation footer"> 279 <tr> 280 <td width="40%" align="left"><a accesskey="p" href="concepts.html">Prev</a> </td> 281 <td width="20%" align="center"> 282 <a accesskey="u" href="introduction.html">Up</a> 283 </td> 284 <td width="40%" align="right"> <a accesskey="n" href="databaseLimits.html">Next</a></td> 285 </tr> 286 <tr> 287 <td width="40%" align="left" valign="top">Berkeley DB Concepts </td> 288 <td width="20%" align="center"> 289 <a accesskey="h" href="index.html">Home</a> 290 </td> 291 <td width="40%" align="right" valign="top"> Database Limits and Portability</td> 292 </tr> 293 </table> 294 </div> 295 </body> 296</html> 297