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 V1.62.4" /> 9 <link rel="home" 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="previous" href="javadplconcepts.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="javadplconcepts.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></div> 36 </div> 37 <p> 38 While this manual will focus primarily on the BTree access method, it is 39 still useful to briefly describe all of the access methods that DB 40 makes available. 41 </p> 42 <div class="note" style="margin-left: 0.5in; margin-right: 0.5in;"> 43 <h3 class="title">Note</h3> 44 <p> 45 If you are using the DPL, be aware that it only 46 supports the BTree access method. For that reason, you 47 can skip this section. 48 </p> 49 </div> 50 <p> 51 Note that an access method can be selected only when the database is 52 created. Once selected, actual API usage is generally 53 identical across all access methods. That is, while some 54 exceptions exist, mechanically you interact with the library in the same 55 way regardless of which access method you have selected. 56 </p> 57 <p> 58 The access method that you should choose is gated first by what you want 59 to use as a key, and then secondly by the performance that you see 60 for a given access method. 61 </p> 62 <p> 63 The following are the available access methods: 64 </p> 65 <div class="informaltable"> 66 <table border="1" width="80%"> 67 <colgroup> 68 <col align="left" /> 69 <col align="left" /> 70 </colgroup> 71 <thead> 72 <tr> 73 <th align="center">Access Method</th> 74 <th align="center">Description</th> 75 </tr> 76 </thead> 77 <tbody> 78 <tr> 79 <td align="left">BTree</td> 80 <td align="left" valign="top"> 81 <p> 82 Data is stored in a sorted, balanced tree structure. 83 Both the key and the data for BTree records can be 84 arbitrarily complex. That is, they can contain single values 85 such as an integer or a string, or complex types such as a 86 structure. Also, although not the default 87 behavior, it is possible for two records to 88 use keys that compare as equals. When this occurs, the 89 records are considered to be duplicates of one another. 90 </p> 91 </td> 92 </tr> 93 <tr> 94 <td align="left">Hash</td> 95 <td align="left" valign="top"> 96 <p> 97 Data is stored in an extended linear hash table. Like 98 BTree, the key and the data used for Hash records can be of 99 arbitrarily complex data. Also, like BTree, duplicate 100 records are optionally supported. 101 </p> 102 </td> 103 </tr> 104 <tr> 105 <td align="left">Queue</td> 106 <td align="left" valign="top"> 107 <p> 108 Data is stored in a queue as fixed-length records. Each 109 record uses a logical record number as its key. This access 110 method is designed for fast inserts at the tail of the 111 queue, and it has a special operation that deletes and 112 returns a record from the head of the queue. 113 </p> 114 <p> 115 This access method is unusual in that it provides record 116 level locking. This can provide 117 beneficial performance improvements in applications 118 requiring concurrent access to the queue. 119 </p> 120 </td> 121 </tr> 122 <tr> 123 <td align="left">Recno</td> 124 <td align="left" valign="top"> 125 <p> 126 Data is stored in either fixed or variable-length records. 127 Like Queue, Recno records use logical record numbers as keys. 128 </p> 129 </td> 130 </tr> 131 </tbody> 132 </table> 133 </div> 134 <div class="sect2" lang="en" xml:lang="en"> 135 <div class="titlepage"> 136 <div> 137 <div> 138 <h3 class="title"><a id="selectAM"></a>Selecting Access Methods</h3> 139 </div> 140 </div> 141 <div></div> 142 </div> 143 <p> 144 To select an access method, you should first consider what you want 145 to use as a key for you database records. If you want to use 146 arbitrary data (even strings), then you should use either BTree or 147 Hash. If you want to use logical record numbers (essentially 148 integers) then you should use Queue or Recno. 149 </p> 150 <p> 151 Once you have made this decision, you must choose between either 152 BTree or Hash, or Queue or Recno. This decision is described next. 153 </p> 154 </div> 155 <div class="sect2" lang="en" xml:lang="en"> 156 <div class="titlepage"> 157 <div> 158 <div> 159 <h3 class="title"><a id="BTreeVSHash"></a>Choosing between BTree and Hash</h3> 160 </div> 161 </div> 162 <div></div> 163 </div> 164 <p> 165 For small working datasets that fit entirely in memory, there is no 166 difference between BTree and Hash. Both will perform just as well 167 as the other. In this situation, you might just as well use BTree, 168 if for no other reason than the majority of DB applications use 169 BTree. 170 </p> 171 <p> 172 Note that the main concern here is your 173 working dataset, not your entire dataset. Many applications maintain 174 large amounts of information but only need to access some small 175 portion of that data with any frequency. So what you want to 176 consider is the data that you will routinely use, not the sum total 177 of all the data managed by your application. 178 </p> 179 <p> 180 However, as your working dataset grows to the point 181 where you cannot fit it all into memory, then you need to take more 182 care when choosing your access method. Specifically, choose: 183 </p> 184 <div class="itemizedlist"> 185 <ul type="disc"> 186 <li> 187 <p> 188 BTree if your keys have some locality of reference. That is, 189 if they sort well and you can expect that a query for a 190 given key will likely be followed by a query for one of its 191 neighbors. 192 </p> 193 </li> 194 <li> 195 <p> 196 Hash if your dataset is extremely large. For any given 197 access method, DB must maintain a certain amount of internal 198 information. However, the amount of information that DB 199 must maintain for BTree is much greater than for Hash. The 200 result is that as your dataset grows, this internal 201 information can dominate the cache to the point where there 202 is relatively little space left for application data. 203 As a result, BTree can be forced to perform disk I/O much more 204 frequently than would Hash given the same amount of data. 205 </p> 206 <p> 207 Moreover, if your dataset becomes so large that DB will 208 almost certainly have to perform disk I/O to satisfy a 209 random request, then Hash will definitely out perform BTree 210 because it has fewer internal records to search through than 211 does BTree. 212 </p> 213 </li> 214 </ul> 215 </div> 216 </div> 217 <div class="sect2" lang="en" xml:lang="en"> 218 <div class="titlepage"> 219 <div> 220 <div> 221 <h3 class="title"><a id="QueueVSRecno"></a>Choosing between Queue and Recno</h3> 222 </div> 223 </div> 224 <div></div> 225 </div> 226 <p> 227 Queue or Recno are used when the application wants to use logical 228 record numbers for the primary database key. Logical record numbers 229 are essentially integers that uniquely identify the database 230 record. They can be either mutable or fixed, where a mutable record 231 number is one that might change as database records are stored or 232 deleted. Fixed logical record numbers never change regardless of 233 what database operations are performed. 234 </p> 235 <p> 236 When deciding between Queue and Recno, choose: 237 </p> 238 <div class="itemizedlist"> 239 <ul type="disc"> 240 <li> 241 <p> 242 Queue if your application requires high degrees of 243 concurrency. Queue provides record-level locking (as opposed 244 to the page-level locking that the other access methods 245 use), and this can result in significantly faster throughput 246 for highly concurrent applications. 247 </p> 248 <p> 249 Note, however, that Queue provides support only for fixed 250 length records. So if the size of the data that you want to 251 store varies widely from record to record, you should 252 probably choose an access method other than Queue. 253 </p> 254 </li> 255 <li> 256 <p> 257 Recno if you want mutable record numbers. Queue is only 258 capable of providing fixed record numbers. Also, Recno 259 provides support for databases whose permanent storage is a 260 flat text file. This is useful for applications looking for 261 fast, temporary storage while the data is being read or 262 modified. 263 </p> 264 </li> 265 </ul> 266 </div> 267 </div> 268 </div> 269 <div class="navfooter"> 270 <hr /> 271 <table width="100%" summary="Navigation footer"> 272 <tr> 273 <td width="40%" align="left"><a accesskey="p" href="javadplconcepts.html">Prev</a>��</td> 274 <td width="20%" align="center"> 275 <a accesskey="u" href="introduction.html">Up</a> 276 </td> 277 <td width="40%" align="right">��<a accesskey="n" href="databaseLimits.html">Next</a></td> 278 </tr> 279 <tr> 280 <td width="40%" align="left" valign="top">Berkeley DB Concepts��</td> 281 <td width="20%" align="center"> 282 <a accesskey="h" href="index.html">Home</a> 283 </td> 284 <td width="40%" align="right" valign="top">��Database Limits and Portability</td> 285 </tr> 286 </table> 287 </div> 288 </body> 289</html> 290