Revision as of 13:10, 16 June 2011

Single-Table Multi-Tenancy

The SINGLE_TABLE multi-tenant type specifies that any table to which an entity or mapped

superclass maps can include rows for multiple tenants. Access to tenant-specific rows is restricted to

the tenant.

Tenant-specific rows are associated with the tenant by using tenant discriminator columns. The

discriminator columns are used with application context values to limit what a persistence context

can access.

The results of queries on the mapped tables are limited to the tenant discriminator value(s) provided as

property values. This applies to all insert, update, and delete operations on the table. When multi-tenant metadata is applied at the mapped superclass level, it is applied to all subentities

unless they specify their own multi-tenant metadata. 'REVIEWERS: I didn’t include any of the info re:

homogeneity from the design doc, because I took it to be implementation details that will be pertinent

when TABLE_PER_TENANT is available. Therefore, I left it out. Please advise if I should add any back

in.'

Note: In the context of single-table multi-tenancy, “single-table” means multiple tenants can share

a single table, and each tenant’s data is distinguished from other tenants’ data via the discriminator

column(s). This is opposed to a different kind of multi-tenancy where each tenant’s data is stored in

separate tables, and the tenant’s data is distinguished from other tenants’ data by identifying the

table. It is possible to use multiple tables with single-table multi-tenancy; but in that case, an

entity’s persisted data is stored in multiple, and multiple tenants can share all the tables.

@Multitenant Attributes

@TenantDiscriminatorColumn Attributes

Configuring Single-Table Multi-Tenancy

To configure single-table multi-tenancy, you must specify both of the following:

1. Annotate the entity or mapped superclass to use single-table multi-tenancy, using the

@Multitenant annotation,

@Multitenant

annotation, for example:

@Entity
@Table(name=“EMP”)
@Multitenant(SINGLE_TABLE)

SINGLE_TABLE states that the table or tables (Table and SecondaryTable) associated with the given entity

can be shared among tenants. 2.      Specify the column or columns to be used as the discriminator column, using the

@TenantDiscriminatorColumn annotation, for example: @Entity @Table(name=“EMP”) @Multitenant(SINGLE_TABLE) @TenantDiscriminatorColumn(name = “TENANT_ID”)   You can specify multiple discriminator columns by using the @TableDiscriminatorColumns annotation, for

example:   @Entity @Table(name = "EMPLOYEE") @Multitenant(SINGLE_TABLE) @TenantDiscriminatorColumns({

   @TenantDiscriminatorColumn(name = "TENANT_ID")
         @TenantDiscriminatorColumn(name = "TENANT_CODE")})

Using Discriminator Columns

The following characteristics apply to discriminator columns:  

•         Tenant discriminator columns are completely application definable. REVIEWERS: I’m not sure what

this sentence is after. Can I rewrite as “Discriminator columns must be defined by the application.”?

•         There is no limit on how many tenant discriminator columns an application can define.
•         Any name can be used for a discriminator column.
•         Tenant discriminator column(s) must always be used with @Multitenant(SINGLE_TABLE). You cannot

specify the tenant discriminator column(s) only.

•         Generated schemas can include specified tenant discriminator columns.
•         Tenant discriminator columns can be mapped or unmapped:

o       When a tenant discriminator column is mapped, its associated mapping attribute must be marked as

read only. With this restriction in place, a tenant discriminator column cannot be part of the entity

identifier; it can only be part of the primary key specification on the database. o       On persist, the value of a mapped tenant discriminator column mapping is populated from its

associated context property. o       Both mapped and unmapped properties are used to form the additional criteria when issuing a

SELECT query. o       When a tenant discriminator column is not mapped, the row is populated with the tenant

discriminator column’s associated context property value. REVIEWERS: I have a question about this item from the design doc: “Unmapped tenant discriminator columns will require EclipseLink to populate the row with the tenant

discriminator column’s associated context property value. See Core section below.” I am confused by this. There is also the item: “On persist, the value of a mapped tenant discriminator column mapping is populated from its associated

context property. “ It sounds like these two might be saying the same thing for mapped and unmapped. What am I missing?

Using Single-Table Multi-Tenancy in an Inheritence Hierarchy

Inheritance strategies are configured by specifying the inheritance type (see @inheritence in

javax.persistence). Single-table multi-tenancy can be used in an inheritance hierarchy, as follows:

• Multi-tenant metadata can only be applied at the root level of the inheritance hierarchy when using a

SINGLE_TABLE or JOINED inheritance strategy.

• It is possible to specify multi-tenant metadata within a TABLE_PER_CLASS inheritance hierarchy.

REVIEWERS: How? Is it that the metadata can be applied somewhere down the hierarchy?

Annotation Examples

The following example defines a single discriminator tenant column:   @Entity @Table(name = "CUSTOMER") @Multitenant @TenantDiscriminatorColumn(name = "TENANT",

    contextProperty = "multi-tenant.id")

public Customer() {

 ...

}   The following example defines multiple tenant discriminator columns using multiple tables:   @Entity @Table(name = "EMPLOYEE") @SecondaryTable(name = "RESPONSIBILITIES") @Multitenant(SINGLE_TABLE) @TenantDiscriminatorColumns({

   @TenantDiscriminatorColumn(name = "TENANT_ID", 
       contextProperty = "employee-tenant.id", 
       length = 20)
   @TenantDiscriminatorColumn(name = "TENANT_CODE", 
       contextProperty = "employee-tenant.code", 
       discriminatorType = STRING, 
       table = "RESPONSIBILITIES")
 }

) public Employee() {

 ...

}   The following example defines a tenant discriminator column mapped as part of the primary key on the

database:   @Entity @Table(name = "ADDRESS") @Multitenant @TenantDiscriminatorColumn(name = "TENANT", contextProperty = "tenant.id",

   primaryKey = true)

public Address() {

 ...

}   The following example defines a mapped tenant discriminator column:   @Entity @Table(name = "Player") @Multitenant @TenantDiscriminatorColumn(name = "AGE",

 contextProperty = "tenant.age")

public Player() {

 ...

@Basic @Column(name="AGE", insertable="false", updatable="false") public int age; }

Specifying Metadata Using XML

You can also use XML to configure single-table multi-tenancy, as an alternative to or in addition to

using annotations. Use the following elements in the eclipselink-orm.xml file: <multitenant> <multitenant-type> <tenant-discriminator-column>   For the EclipseLink schemas, see http://wiki.eclipse.org/EclipseLink/XSDs. REVIEWERS: The design doc makes it look like you don’t have to specify <multitenant-type> when using XML.

Is that correct? XML Examples The following example defines a single tenant discriminator column: <entity class="model.Customer">

 <multitenant>
   <tenant-discriminator-column 
     name="TENANT” 

context-property="multi-tenant.id"/>

 </multitenant>
 <table name="CUSTOMER"/>
 ...

</entity>     The following example defines multiple tenant discriminator columns using multiple tables:   <entity class="model.Employee">

 <multitenant type="SINGLE_TABLE">
   <tenant-discriminator-column name="TENANT_ID" context-property="employee-tenant.id" length="20"/>
   <tenant-discriminator-column name="TENANT_CODE" context-property="employee-tenant.id" 

discriminator-type="STRING" table="RESPONSIBILITIES"/>

 </multitenant>
 <table name="EMPLOYEE"/>
 <secondary-table name="RESPONSIBILITIES"/>
 ...

</entity>   The following example defines a tenant discriminator column mapped as part of the primary key on the

database:   <entity class="model.Address">

 <multitenant>
   <tenant-discriminator-column name="TENANT" context-property="multi-tenant.id" primary-key="true"/>
 </multitenant>
 <table name="ADDRESS"/>
 ...

</entity>   The following example defines a mapped tenant discriminator column:   <entity class="model.Player">

 <multi-tenant>
   <tenant-discriminator-column name="AGE" context-property="tenant.age"/>
 </multi-tenant>
 <table name="PLAYER"/>
 ...
 <attributes>
   <basic name="age" insertable="false" updatable="false">
     <column name="AGE"/>
   </basic>
   ...
 </attributes>
 ...

</entity>  

Property Configuration and Caching Scope

At runtime, context properties can be specified in a persistence unit definition or passed to a

CreateEntityManagerFactory() call, as shown in the following examples.

The order of precedence for tenant discriminator column properties is as follows, listed from highest

priority to lowest: <-- REVIEWERS: is that right? Highest to lowest?

1. EntityManager
2. EntityManagerFactory
3. Application context (when in a Java EE container)

The following example shows a property set in a persistence unit definition in the persistence.xml file:   <persistence-unit name="multi-tenant">

 ...
 <properties>
   <property name="tenant.id" value="707"/>
   ...
 </properties>

</persistence-unit>   The following example shows the property set in code:   HashMap properties = new HashMap(); properties.put("tenant.id", "707"); ... EntityManager em = Persistence.createEntityManagerFactory("multi-tenant",

properties).createEntityManager();

Entity Manager Factory

At the level of the entity manager factory, you must provide a unique session name through the

eclipselink.session-name property, to ensure a unique server session (and cache) is provided for each

tenant. This allows for user-defined properties (without any prefixing). For example:   HashMap properties = new HashMap(); properties.put("tenant.id", "707"); properties.put("eclipselink.session-name", "multi-tenant-707"); ... EntityManager em = Persistence.createEntityManagerFactory("multi-tenant",

properties).createEntityManager();  

Shared Entity Manager Factory

When using a shared entity manager factory, you must set following property to indicate the factory will

be shared: eclipselink.multitenant.tenants-share-cache

When this property is set, all multitenant entities will have a PROTECTED cache setting. Entity Manager

At the level of the entity manager, you must specify caching strategies, because the same server session

can be used for each tenant. For example, you can use an isolation level to ensure no shared tenant

information exists in the L2 cache. These settings are set when creating the entity manager factory. For

example: HashMap tenantProperties = new HashMap(); properties.put("tenant.id", "707");   HashMap cacheProperties = new HashMap(); properties.put("eclipselink.cache.shared.Employee", "false"); properties.put("eclipselink.cache.size.Address", "10"); properties.put("eclipselink.cache.type.Contract", "NONE"); ... EntityManager em = Persistence.createEntityManagerFactory("multi-tenant",

cacheProperties).createEntityManager(tenantProperties); ... Swapping tenant ID during a live EntityManager is not allowed.

Defining Persistence Unit and Entity Mappings Defaults

You can define single-table multi-tenancy for specific entities and mapped superclasses, as described

above. In addition, you can define the metadata at higher levels in the eclipselink-orm.xml file, to

provide defaults. Standard JPA metadata defaulting and overriding rules apply. The elements used for

specifying these defaults are:

• <persistence-unit-defaults>
• <entity-mappings>

Persistence Unit Defaults

You can specify default tenant discriminator column metadata using the <persistence-unit-defaults>

element and its <tenant-discriminator-column> subelement. When defined at this level, it applies to all

entities of the persistence unit that specify the SINGLE_TABLE multi-tenant type, excluding those that

specify their own tenant discriminator metadata. For example: REVIWERS: I came up with this example myself. Is it correct? If not, please advise.

<persistence-unit-metadata>

 <persistence-unit-defaults>
 ...
   <tenant-discriminator-column 
     name="TENANT” 

context-property="multi-tenant.id"/>

   </multitenant>
 </persistence-unit-defaults>

</persistence-unit-metadata>   Note: With no defaults, an entity not marked with multi-tenant metadata will not use any multi-tenancy

strategy.

Entity Mappings Defaults

You can also specify tenant discriminator column metadata using <entity-mappings> element. This overrides

persistence unit defaults (described above) and applies to all entities of the given mapping file that

specify the SINGLE_TABLE multi-tenant type, excluding those entities that specify their own tenant

discriminator metadata. For example, REVIWERS: I came up with this example myself. Is it correct? If not, please advise.

<entity-mappings >

 ...
   <tenant-discriminator-column 
     name="TENANT” 

context-property="multi-tenant.id"/> <entity-mappings >

Example

Defaults always apply even when there are multiple tenant discriminators. This allows you to map several

columns for the same property. For example, the code below defaults the property name to

"eclipselink.tenant-id" and states it should be writing the TENANT column for both the EMPLOYEE and

SALARY table. @Entity @Table(name = "EMPLOYEE") @SecondaryTable(name = "SALARY") @MultiTenant(SINGLE_TABLE) @TenantDiscriminatorColumns({

   @TenantDiscriminatorColumn(name = "TENANT")
   @TenantDiscriminatorColumn(name = "TENANT", table = "SALARY")
 }

) public Employee() {

 ...

}  

Public Default Context Property

The public default context property (eclipselink.tenant-id) definition is available from:

• org.eclipse.persistence.config.EntityManagerProperties.MULTITENANT_PROPERTY_DEFAULT
• org.eclipse.persistence.config.PersistenceUnitProperties.MULTITENANT_PROPERTY_DEFAULT

For example:

EntityManager em = createEntityManager(MULTI_TENANT_PU); em.setProperty("tenant.id", "707"); em.setProperty(EntityManagerProperties.MULTITENANT_PROPERTY_DEFAULT, "707");

Support for Tenant Discriminator Columns through Entity Manager Operations and Querying The tenant discriminator column and value are supported through the following entity manager operations:

• persist()
• find()
• refresh()

They are also supported through the following queries:

• Named queries
• Update all
• Delete all

REVIEWERS: Do you think any examples would be appropriate here? If so, please provide. Note: EclipseLink does not support multi-tenancy through named native queries. If you want to use named

native queries in a multi-tenant environment, you must handle any multi-tenancy issues directly in the

query. In general, it is best to avoid named native queries in a multi-tenant environment.