<?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>
The contents of this file are subject to the Erlang Public License,
Version 1.1, (the "License"); you may not use this file except in
compliance with the License. You should have received a copy of the
Erlang Public License along with this software. If not, it can be
retrieved online at http://www.erlang.org/.
Software distributed under the License is distributed on an "AS IS"
basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See
the License for the specific language governing rights and limitations
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>
