<?xml version="1.0" encoding="UTF-8" standalone="no"?> <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> <html xmlns="http://www.w3.org/1999/xhtml"> <head> <meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /> <title>Reverse BTree Splits</title> <link rel="stylesheet" href="gettingStarted.css" type="text/css" /> <meta name="generator" content="DocBook XSL Stylesheets V1.62.4" /> <link rel="home" href="index.html" title="Getting Started with Berkeley DB Transaction Processing" /> <link rel="up" href="txnconcurrency.html" title="Chapter 4. Concurrency" /> <link rel="previous" href="txnnowait.html" title="No Wait on Blocks" /> <link rel="next" href="filemanagement.html" title="Chapter 5. Managing DB Files" /> </head> <body> <div class="navheader"> <table width="100%" summary="Navigation header"> <tr> <th colspan="3" align="center">Reverse BTree Splits</th> </tr> <tr> <td width="20%" align="left"><a accesskey="p" href="txnnowait.html">Prev</a> </td> <th width="60%" align="center">Chapter 4. Concurrency</th> <td width="20%" align="right"> <a accesskey="n" href="filemanagement.html">Next</a></td> </tr> </table> <hr /> </div> <div class="sect1" lang="en" xml:lang="en"> <div class="titlepage"> <div> <div> <h2 class="title" style="clear: both"><a id="reversesplit"></a>Reverse BTree Splits</h2> </div> </div> <div></div> </div> <p> If your application is using the Btree access method, and your application is repeatedly deleting then adding records to your database, then you might be able to reduce lock contention by turning off reverse Btree splits. </p> <p> As pages are emptied in a database, DB attempts to delete empty pages in order to keep the database as small as possible and minimize search time. Moreover, when a page in the database fills up, DB, of course, adds additional pages to make room for more data. </p> <p> Adding and deleting pages in the database requires that the writing thread lock the parent page. Consequently, as the number of pages in your database diminishes, your application will see increasingly more lock contention; the maximum level of concurrency in a database of two pages is far smaller than that in a database of 100 pages, because there are fewer pages that can be locked. </p> <p> Therefore, if you prevent the database from being reduced to a minimum number of pages, you can improve your application's concurrency throughput. Note, however, that you should do so only if your application tends to delete and then add the same data. If this is not the case, then preventing reverse Btree splits can harm your database search time. </p> <p> To turn off reverse Btree splits, <span> provide the <tt class="literal">DB_REVSPLITOFF</tt> flag to the <tt class="methodname">Db::set_flags()</tt> method. </span> </p> <p> For example: </p> <pre class="programlisting">#include "db_cxx.h" ... int main(void) { u_int32_t env_flags = DB_CREATE | // If the environment does not // exist, create it. DB_INIT_LOCK | // Initialize locking DB_INIT_LOG | // Initialize locking DB_INIT_MPOOL | // Initialize the cache DB_THREAD | // Free-thread the env handle DB_INIT_TXN; // Initialize transactions u_int32_t db_flags = DB_CREATE | DB_AUTO_COMMIT; Db *dbp = NULL; const char *file_name = "mydb.db"; std::string envHome("/export1/testEnv"); DbEnv myEnv(0); try { myEnv.open(envHome.c_str(), env_flags, 0); dbp = new Db(&myEnv, 0); // Turn off BTree reverse split. dbp=>set_flags(DB_REVSPLITOFF); dbp->open(dbp, // Pointer to the database NULL, // Txn pointer file_name, // File name NULL, // Logical db name DB_BTREE, // Database type (using btree) db_flags, // Open flags 0); // File mode. Using defaults } catch(DbException &e) { std::cerr << "Error opening database and environment: " << file_name << ", " << envHome << std::endl; std::cerr << e.what() << std::endl; } try { dbp->close(dbp, 0); myEnv.close(0); } catch(DbException &e) { std::cerr << "Error closing database and environment: " << file_name << ", " << envHome << std::endl; std::cerr << e.what() << std::endl; return (EXIT_FAILURE); } return (EXIT_SUCCESS); } </pre> </div> <div class="navfooter"> <hr /> <table width="100%" summary="Navigation footer"> <tr> <td width="40%" align="left"><a accesskey="p" href="txnnowait.html">Prev</a> </td> <td width="20%" align="center"> <a accesskey="u" href="txnconcurrency.html">Up</a> </td> <td width="40%" align="right"> <a accesskey="n" href="filemanagement.html">Next</a></td> </tr> <tr> <td width="40%" align="left" valign="top">No Wait on Blocks </td> <td width="20%" align="center"> <a accesskey="h" href="index.html">Home</a> </td> <td width="40%" align="right" valign="top"> Chapter 5. Managing DB Files</td> </tr> </table> </div> </body> </html>