<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE chapter SYSTEM "chapter.dtd">

<chapter>
  <header>
    <copyright>
      <year>1997</year><year>2013</year>
      <holder>Ericsson AB. All Rights Reserved.</holder>
    </copyright>
    <legalnotice>
      Licensed under the Apache License, Version 2.0 (the "License");
      you may not use this file except in compliance with the License.
      You may obtain a copy of the License at
 
          http://www.apache.org/licenses/LICENSE-2.0

      Unless required by applicable law or agreed to in writing, software
      distributed under the License is distributed on an "AS IS" BASIS,
      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
      See the License for the specific language governing permissions and
      limitations under the License.
      under the License.

    </legalnotice>

    <title>Build a Mnesia Database</title>
    <prepared></prepared>
    <responsible></responsible>
    <docno></docno>
    <approved></approved>
    <checked></checked>
    <date></date>
    <rev></rev>
    <file>Mnesia_chap3.xml</file>
  </header>
  <p>This section describes the basic steps when designing a
    <c>Mnesia</c> database and the programming constructs that make different
    solutions available to the programmer. The following topics are
    included:</p>
  <list type="bulleted">
    <item>Define a schema</item>
    <item>Data model</item>
    <item>Start <c>Mnesia</c></item>
    <item>Create tables</item>
  </list>

  <section>
    <marker id="def_schema"></marker>
    <title>Define a Schema</title>
    <p>The configuration of a <c>Mnesia</c> system is described in a
      schema. The schema is a special table that includes information
      such as the table names and the storage type of each table
      (that is, whether a table is to be stored in RAM,
      on disc, or on both, as well as its location).</p>
    <p>Unlike data tables, information in schema tables can only be
      accessed and modified by using the schema-related functions
      described in this section.</p>
    <p><c>Mnesia</c> has various functions for defining the
      database schema. Tables can be moved or deleted, and the
      table layout can be reconfigured.</p>
    <p>An important aspect of these functions is that the system can access
      a table while it is being reconfigured. For example, it is possible
      to move a
      table and simultaneously perform write operations to the same
      table. This feature is essential for applications that require
      continuous service.</p>
    <p>This section describes the functions available for schema management,
      all which return either of the following tuples:</p>
    <list type="bulleted">
      <item><c>{atomic, ok}</c> if successful</item>
      <item><c>{aborted, Reason}</c> if unsuccessful</item>
    </list>

    <section>
      <title>Schema Functions</title>
    <p>The schema functions are as follows:</p>
      <list type="bulleted">
        <item><seealso marker="mnesia#create_schema/1">mnesia:create_schema(NodeList)</seealso>
         initializes a new, empty schema. This is a mandatory requirement
         before <c>Mnesia</c> can be started. <c>Mnesia</c> is a truly
         distributed DBMS and the schema is a system table that is
         replicated on all nodes in a <c>Mnesia</c> system.
         This function fails if a schema is already present on any of
         the nodes in <c>NodeList</c>. The function requires <c>Mnesia</c>
         to be stopped on the all
         <c>db_nodes</c> contained in parameter <c>NodeList</c>.
         Applications call this function only once, as
         it is usually a one-time activity to initialize a new database.
        </item>
        <item><seealso marker="mnesia#delete_schema/1">mnesia:delete_schema(DiscNodeList)</seealso>
         erases any old schemas on the nodes in
         <c>DiscNodeList</c>. It also removes all old tables together
         with all data. This function requires <c>Mnesia</c> to be stopped
         on all <c>db_nodes</c>.
        </item>
        <item><seealso marker="mnesia#delete_table/1">mnesia:delete_table(Tab)</seealso>
         permanently deletes all replicas of table <c>Tab</c>.
        </item>
        <item><seealso marker="mnesia#clear_table/1">mnesia:clear_table(Tab)</seealso>
         permanently deletes all entries in table <c>Tab</c>.
        </item>
        <item><seealso marker="mnesia#move_table_copy/3">mnesia:move_table_copy(Tab, From, To)</seealso>
         moves the copy of table <c>Tab</c> from node
         <c>From</c> to node <c>To</c>. The table storage type
         <c>{type}</c> is preserved, so if a RAM table is moved from
         one node to another, it remains a RAM table on the new
         node. Other transactions can still perform read
         and write operation to the table while it is being moved.
        </item>
        <item><seealso marker="mnesia#add_table_copy/3">mnesia:add_table_copy(Tab, Node, Type)</seealso>
         creates a replica of table <c>Tab</c> at node
         <c>Node</c>. Argument <c>Type</c> must be either of the
         atoms <c>ram_copies</c>, <c>disc_copies</c>, or
         <c>disc_only_copies</c>. If you add a copy of the system
         table <c>schema</c> to a node, you want the <c>Mnesia</c>
         schema to reside there as well. This action
         extends the set of nodes that comprise this particular
         <c>Mnesia</c> system.
        </item>
        <item><seealso marker="mnesia#del_table_copy/2">mnesia:del_table_copy(Tab, Node)</seealso>
         deletes the replica of table <c>Tab</c> at node <c>Node</c>.
         When the last replica of a table is removed, the table is
         deleted.
        </item>
        <item>
          <p><seealso marker="mnesia#transform_table/4">mnesia:transform_table(Tab, Fun, NewAttributeList, NewRecordName)</seealso>
            changes the format on all records in table
            <c>Tab</c>. It applies argument <c>Fun</c> to all
            records in the table. <c>Fun</c> must be a function that
            takes a record of the old type, and returns the record of the
            new type. The table key must not be changed.</p>
          <p><em>Example:</em></p>
          <code type="none">
-record(old, {key, val}).
-record(new, {key, val, extra}).

Transformer =
   fun(X) when record(X, old) ->
      #new{key = X#old.key,
           val = X#old.val,
           extra = 42}
   end,
{atomic, ok} = mnesia:transform_table(foo, Transformer,
                                      record_info(fields, new),
                                      new),
</code>
            <p>Argument <c>Fun</c> can also be the atom
            <c>ignore</c>, which indicates that only the metadata about
            the table is updated. Use of <c>ignore</c> is not recommended
            (as it creates inconsistencies between the metadata and the
            actual data) but it is included as a possibility for the user
            do to an own (offline) transform.</p>
        </item>
        <item><c>change_table_copy_type(Tab, Node, ToType)</c>
         changes the storage type of a table. For example, a
         RAM table is changed to a <c>disc_table</c> at the node specified
         as <c>Node</c>.</item>
      </list>
    </section>
  </section>

  <section>
    <title>Data Model</title>
    <p>The data model employed by <c>Mnesia</c> is an extended
      relational data model. Data is organized as a set of
      tables and relations between different data records can
      be modeled as more tables describing the relationships.
      Each table contains instances of Erlang records.
      The records are represented as Erlang tuples.</p>
    <p>Each Object Identifier (OID) is made up of a table name and a key.
      For example, if an employee record is represented by the tuple
      <c>{employee, 104732, klacke, 7, male, 98108, {221, 015}}</c>,
      this record has an OID, which is the tuple
      <c>{employee, 104732}</c>.</p>
    <p>Thus, each table is made up of records, where the first element
      is a record name and the second element of the table is a key,
      which identifies the particular record in that table. The
      combination of the table name and a key is an arity two tuple
      <c>{Tab, Key}</c> called the OID. For more information about
      the relationship beween the record name and the table name, see
      <seealso marker="Mnesia_chap4#recordnames_tablenames">Record Names versus Table Names</seealso>.
    </p>
    <p>What makes the <c>Mnesia</c> data model an extended relational model
      is the ability to store arbitrary Erlang terms in the attribute
      fields. One attribute value can, for example, be a whole tree of
      OIDs leading to other terms in other tables. This type
      of record is difficult to model in traditional relational DBMSs.</p>
  </section>

  <section>
    <marker id="start_mnesia"></marker>
    <title>Start Mnesia</title>
    <p>Before starting <c>Mnesia</c>, the following must be done:
      </p>
    <list type="bulleted">
      <item>An empty schema must be initialized on all the
       participating nodes.</item>
      <item>The Erlang system must be started.</item>
      <item>Nodes with disc database schema must be defined and
       implemented with the function
       <seealso marker="mnesia#create_schema/1">mnesia:create_schema(NodeList)</seealso>.</item>
    </list>
    <p>When running a distributed system with two or more
      participating nodes, the function
      <seealso marker="mnesia#start/0">mnesia:start()</seealso>
      must be executed on each participating node. This would typically
      be part of the boot script in an embedded environment.
      In a test environment or an interactive environment,
      <c>mnesia:start()</c> can also be used either from the
      Erlang shell or another program.
      </p>

    <section>
      <title>Initialize a Schema and Start Mnesia</title>
      <p>Let us use the example database <c>Company</c>, described in
        <seealso marker="Mnesia_chap2#getting_started">Getting Started</seealso> to
        illustrate how to run a database on two separate nodes,
        called <c>a@gin</c> and <c>b@skeppet</c>. Each of these
        nodes must have a <c>Mnesia</c> directory and an
        initialized schema before <c>Mnesia</c> can be started. There are
        two ways to specify the <c>Mnesia</c> directory to be used:</p>
      <list type="bulleted">
        <item>
          <p>Specify the <c>Mnesia</c> directory by providing an application
            parameter either when starting the Erlang shell or in the
            application script. Previously, the following example was used
            to create the directory for the <c>Company</c> database:</p>
          <pre>
%<input>erl -mnesia dir '"/ldisc/scratch/Mnesia.Company"'</input>
          </pre>
        </item>
        <item>If no command-line flag is entered, the <c>Mnesia</c>
         directory becomes the current working directory on the node
         where the Erlang shell is started.</item>
      </list>
      <p>To start the <c>Company</c> database and get it running on the two
        specified nodes, enter the following commands:</p>
      <list type="ordered">
        <item>
          <p>On the node <c>a@gin</c>:</p>
          <pre>
 gin %<input>erl -sname a  -mnesia dir '"/ldisc/scratch/Mnesia.company"'</input></pre>
        </item>
        <item><p>On the node <c>b@skeppet</c>:</p>
          <pre>
skeppet %<input>erl -sname b -mnesia dir '"/ldisc/scratch/Mnesia.company"'</input></pre>
        </item>
        <item>
          <p>On one of the two nodes:</p>
          <pre>
(a@gin)1><input>mnesia:create_schema([a@gin, b@skeppet]).</input></pre>
        </item>
        <item>The function
          <seealso marker="mnesia#start/0">mnesia:start()</seealso>
          is called on both nodes.
        </item>
        <item><p>To initialize the database, execute the following
          code on one of the two nodes:</p>
          <codeinclude file="company.erl" tag="%12" type="erl"></codeinclude>
        </item>
      </list>
      <p>As illustrated, the two directories reside on different nodes,
        because <c>/ldisc/scratch</c> (the "local" disc) exists on
        the two different nodes.</p>
      <p>By executing these commands, two Erlang nodes are configured to
        run the <c>Company</c> database, and therefore, initialize the
        database. This is required only once when setting up. The next time
        the system is started,
        <seealso marker="mnesia#start/0">mnesia:start()</seealso>
        is called
        on both nodes, to initialize the system from disc.</p>
      <p>In a system of <c>Mnesia</c> nodes, every node is aware of the
        current location of all tables. In this example, data is
        replicated on both nodes and functions that manipulate the
        data in the tables can be executed on either of the two nodes.
        Code that manipulate <c>Mnesia</c> data behaves identically
        regardless of where the data resides.</p>
      <p>The function <seealso marker="mnesia#stop/0">mnesia:stop()</seealso>
        stops <c>Mnesia</c> on the node
        where the function is executed. The functions <c>mnesia:start/0</c>
        and <c>mnesia:stop/0</c> work on the "local" <c>Mnesia</c> system.
        No functions start or stop a set of nodes.</p>
    </section>

    <section>
      <title>Startup Procedure</title>
      <p>Start <c>Mnesia</c> by calling the following function:</p>
      <code type="none">
          mnesia:start().</code>
      <p>This function initiates the DBMS locally.</p>
      <p>The choice of configuration alters the location and load
        order of the tables. The alternatives are as follows:</p>
      <list type="ordered">
        <item>Tables that are only stored locally are initialized
         from the local <c>Mnesia</c> directory.
        </item>
        <item>Replicated tables that reside locally
         as well as somewhere else are either initiated from disc or
         by copying the entire table from the other node, depending on
         which of the different replicas are the most recent. <c>Mnesia</c>
         determines which of the tables are the most recent.
        </item>
        <item>Tables that reside on remote nodes are available to other
         nodes as soon as they are loaded.</item>
      </list>
      <p>Table initialization is asynchronous. The function
        call <seealso marker="mnesia#start/0">mnesia:start()</seealso>
        returns the atom <c>ok</c> and
        then starts to initialize the different tables. Depending on
        the size of the database, this can take some time, and the
        application programmer must wait for the tables that the
        application needs before they can be used. This is achieved by
        using the function
        <seealso marker="mnesia#wait_for_tables/2">mnesia:wait_for_tables(TabList, Timeout)</seealso>,
        which suspends the caller until all tables
        specified in <c>TabList</c> are properly initiated.</p>
      <p>A problem can arise if a replicated table on one node is
        initiated, but <c>Mnesia</c> deduces that another (remote)
        replica is more recent than the replica existing on the
        local node, and the initialization procedure does not proceed.
        In this situation, a call to
        <seealso marker="mnesia#wait_for_tables/2">mnesia:wait_for_tables/2</seealso>,
        suspends the caller until the
        remote node has initialized the table from its local disc and
        the node has copied the table over the network to the local node.</p>
      <p>However, this procedure can be time-consuming, the shortcut function
        <seealso marker="mnesia#force_load_table/1">mnesia:force_load_table(Tab)</seealso>
        loads all the tables from disc at a faster rate. The function forces
        tables to be loaded from disc regardless of the network
        situation.</p>
      <p>Thus, it can be assumed that if an application wants to use
        tables <c>a</c> and <c>b</c>, the application must perform
        some action similar to following before it can use the tables:</p>
      <pre>
          case mnesia:wait_for_tables([a, b], 20000) of
            {timeout,   RemainingTabs} ->
              panic(RemainingTabs);
            ok ->
              synced
          end.</pre>
      <warning>
        <p>When tables are forcefully loaded from the local disc,
          all operations that were performed on the replicated table
          while the local node was down, and the remote replica was
          alive, are lost. This can cause the database to become
          inconsistent.</p>
      </warning>
      <p>If the startup procedure fails, the function
        <seealso marker="mnesia#start/0">mnesia:start()</seealso>
        returns the cryptic tuple
        <c>{error,{shutdown, {mnesia_sup,start,[normal,[]]}}}</c>.
        To get more information about the start failure, use
        command-line arguments <c>-boot start_sasl</c> as argument to
        the <c>erl</c> script.</p>
    </section>
  </section>

  <section>
    <marker id="create_tables"></marker>
    <title>Create Tables</title>
    <p>The function
      <seealso marker="mnesia#create_table/2">mnesia:create_table(Name, ArgList)</seealso>
      creates tables. When executing this function, it returns one of
      the following responses:</p>
    <list type="bulleted">
      <item><c>{atomic, ok}</c> if the function executes successfully
      </item>
      <item><c>{aborted, Reason}</c> if the function fails
      </item>
    </list>
    <p>The function arguments are as follows:</p>
    <list type="bulleted">
      <item><c>Name</c> is the name of the table. It is
       usually the same name as the name of the records that
       constitute the table. For details, see <c>record_name</c>.
      </item>
      <item>
        <p><c>ArgList</c> is a list of <c>{Key,Value}</c> tuples.
          The following arguments are valid:</p>
        <list type="bulleted">
          <item>
            <p><c>{type, Type}</c>, where <c>Type</c> must be either of
              the atoms <c>set</c>, <c>ordered_set</c>, or <c>bag</c>.
              Default is <c>set</c>.</p>
            <p>Notice that currently <c>ordered_set</c> is not
              supported for <c>disc_only_copies</c> tables.</p>
            <p>A table of type
              <c>set</c> or <c>ordered_set</c> has either zero or
              one record per key, whereas a table of type <c>bag</c> can
              have an arbitrary number of records per key. The key for
              each record is always the first attribute of the record.</p>
            <p>The following example illustrates the difference between
              type <c>set</c> and <c>bag</c>:</p>
            <pre>
 f() ->
    F = fun() ->
          mnesia:write({foo, 1, 2}),
          mnesia:write({foo, 1, 3}),
          mnesia:read({foo, 1})
        end,
    mnesia:transaction(F).</pre>
            <p>This transaction returns the list <c>[{foo,1,3}]</c> if
              table <c>foo</c> is of type <c>set</c>. However, the list
              <c>[{foo,1,2}, {foo,1,3}]</c> is returned if the table is
              of type <c>bag</c>.</p>
            <p><c>Mnesia</c> tables can never contain
              duplicates of the same record in the same table. Duplicate
              records have attributes with the same contents and key.</p>
          </item>
          <item>
            <p><c>{disc_copies, NodeList}</c>, where <c>NodeList</c> is a
              list of the nodes where this table is to reside on disc.</p>
            <p>Write operations to a table replica of type
              <c>disc_copies</c> write data to the disc copy and
              to the RAM copy of the table.</p>
            <p>It is possible to have a
              replicated table of type <c>disc_copies</c> on one node, and
              the same table stored as a different type on another node.
              Default is <c>[]</c>. This arrangement is
              desirable if the following operational
              characteristics are required:</p>
            <list type="ordered">
              <item>Read operations must be fast and performed in RAM.</item>
              <item>All write operations must be written to persistent
               storage.</item>
            </list>
            <p>A write operation on a <c>disc_copies</c> table
              replica is performed in two steps. First the write
              operation is appended to a log file, then the actual
              operation is performed in RAM.</p>
          </item>
          <item>
            <p><c>{ram_copies, NodeList}</c>, where <c>NodeList</c> is a
              list of the nodes where this table is stored in RAM.
              Default is <c>[node()]</c>. If the default value is used
              to create a table, it is located on the local node only.</p>
            <p>Table replicas of type
              <c>ram_copies</c> can be dumped to disc with the function
              <seealso marker="mnesia#dump_tables/1">mnesia:dump_tables(TabList)</seealso>.</p>
          </item>
          <item><c>{disc_only_copies, NodeList}</c>. These table
           replicas are stored on disc only and are therefore slower to
           access. However, a disc-only replica consumes less memory than
           a table replica of the other two storage types.
          </item>
          <item><p><c>{index, AttributeNameList}</c>, where
           <c>AttributeNameList</c> is a list of atoms specifying the
           names of the attributes <c>Mnesia</c> is to build and maintain.
           An index table exists for every element in the list. The first
           field of a <c>Mnesia</c> record is the key and thus need no
           extra index.</p>
           <p>The first field of a record is the second element of the
           tuple, which is the representation of the record.</p>
          </item>
          <item><p><c>{snmp, SnmpStruct}</c>. <c>SnmpStruct</c> is
           described in the
           <seealso marker="snmp:index">SNMP</seealso> User's Guide.
           Basically, if this attribute is present in <c>ArgList</c> of
           <seealso marker="mnesia#create_table/2">mnesia:create_table/2</seealso>,
           the table is immediately accessible the SNMP.</p>
           <p>It is easy to design applications that use SNMP to
           manipulate and control the system. <c>Mnesia</c> provides a
           direct mapping between the logical tables that make up an SNMP
           control application and the physical data that makes up a
           <c>Mnesia</c> table. The default value is <c>[]</c>.</p>
          </item>
          <item><c>{local_content, true}</c>. When an application needs a
           table whose contents is to be locally unique on each node,
           <c>local_content</c> tables can be used. The name of the
           table is known to all <c>Mnesia</c> nodes, but its contents is
           unique for each node. Access to this type of table must be
           done locally.</item>
          <item>
            <p><c>{attributes, AtomList}</c> is a list of the attribute
              names for the records that are supposed to populate the
              table. Default is the list <c>[key, val]</c>. The
              table must at least have one extra attribute besides the
              key. When accessing single attributes in a record, it is not
              recommended to hard code the attribute names as atoms. Use
              the construct <c>record_info(fields, record_name)</c>
              instead.</p>
            <p>The expression
              <c>record_info(fields, record_name)</c> is processed by the
              Erlang preprocessor and returns a list of the
              record field names. With the record definition
              <c>-record(foo, {x,y,z}).</c>, the expression
              <c>record_info(fields,foo)</c> is expanded to the list
              <c>[x,y,z]</c>. It is therefore possible for you to provide
              the attribute names or to use the <c>record_info/2</c>
              notation.</p>
            <p>It is recommended to use the <c>record_info/2</c> notation,
              as it becomes easier to maintain the program and the program
              becomes more robust with regards to future record changes.</p>
          </item>
          <item>
            <p><c>{record_name, Atom}</c> specifies the common name of
              all records stored in the table. All records stored in
              the table must have this name as their first element.
              <c>record_name</c> defaults to the name of the table.
              For more information, see
              <seealso marker="Mnesia_chap4#recordnames_tablenames">Record Names versus Table Names</seealso>.</p>
          </item>
        </list>
      </item>
    </list>
    <p>As an example, consider the following record definition:</p>
    <pre>
      -record(funky, {x, y}).</pre>
    <p>The following call would create a table that is replicated on two
      nodes, has an extra index on attribute <c>y</c>, and is of type
      <c>bag</c>.</p>
    <pre>
      mnesia:create_table(funky, [{disc_copies, [N1, N2]}, {index,
      [y]}, {type, bag}, {attributes, record_info(fields, funky)}]).</pre>
    <p>Whereas a call to the following default code values would return
      a table with a RAM copy on the local node, no extra indexes, and the
      attributes defaulted to the list <c>[key,val]</c>.</p>
    <pre>
mnesia:create_table(stuff, [])</pre>
  </section>
</chapter>