Multi-Tenancy with the SaaS based model does not bring any change on the application layer as multiple services deployment of application would be able to accomodate the load on application but it would matter how do we decide the modelling of the database w.r.t. Multi-Tenancy.
Cost Cost per tenants if seprate DB/SCHEMA is selected per tenants. Performance Application performance & scalability would impact based on the number of tenants, data size per tenants and the workload. Time Development & Release migration effots when multiple databases/schemas are in production. Complexity for managing the operations for maitaining multiple DBs.
Tenancy Modelling
Single Tenant Architecture where each tenant has its own database and is isolated from other tenants.
Multi Tenant Architecture with separate schema where all tenants share a database, but have their own database schemas and tables.
Multi Tenant Architecture with all tenants share a database and tables. Every table has a Column with the Tenant Identifier, that shows the tenant of the row.
Various multitenancy strategy supported by Hibernate are
public enum MultiTenancyStrategy {
DISCRIMINATOR,
SCHEMA,
DATABASE,
NONE;
}
DISCRIMINATOR strartegy work in progress and from the docuemntaiton of hibernate
DISCRIMINATOR strartegy Correlates to the partitioned (discriminator) approach. It is an error to attempt to open a session without a tenant identifier using this strategy. This strategy is not yet implemented and you can follow its progress via the HHH-6054 Jira issue.
JIRA HHH-6054 is currently in Open Status
Here in the example following modules/projects shows the example to achieve the tenancy modelling using SCHEMA (shared db, separate schema) strategy per tenant based approach and SCHEMA (shared db, shared schema) strategy for different tenants.
multitenancy-op2 project shows the example with the multiple tenants share the same database but different schemas.
multitenancy-op3 project shows the example of the hybrid approach of with the tenants are grouped againsts the specific database and schema. This would give an ability to map a specific tenant with single database as well when the high load is expected for a tenant in comparision to others.
UseCase is where database is same (dbA) and every TENANT is using specific Schema i.e.
Tenant wise DB and Schema mapping.
| TENANTID | DB | SCHEMA |
|---|---|---|
| TENANT1 | dbA | SCHEMA1 |
| TENANT2 | dbA | SCHEMA2 |
cd multitenancy-op2
mvn clean package
mvn spring-boot:run
Fly integration would be creating necessary tables in database dbA
Catalog creation request for TENANT1 which would insert data into table SCHEMA1.CATALOGTABLE
curl -X POST http://localhost:8080/catalog -H 'content-type: application/json' -H 'tenantid: TENANT1' -d '{"tenantId":"TENANT1","catalogName":"Catalog1","supplierId":"supplier1","source":"X-Systems"}'
Catalog creation request for TENANT2 which would insert data into table SCHEMA2.CATALOGTABLE
curl -X POST http://localhost:8080/catalog -H 'content-type: application/json' -H 'tenantid: TENANT2' -d '{"tenantId":"TENANT2","catalogName":"Catalog2","supplierId":"supplier2","source":"Y-Systems"}'
Catalog retrival reqeust for TENANT1
curl -X GET 'http://localhost:8080/catalog?catalogId=1' -H 'content-type: application/json' -H 'tenantid: TENANT1'
Catalog retrival reqeust for TENANT2
curl -X GET 'http://localhost:8080/catalog?catalogId=2' -H 'content-type: application/json' -H 'tenantid: TENANT2'
UseCase is as the hybrid approach to accomodate following cases:
- Tenant(s) with higher load using the dedicated database and single schema (TENANT1).
- Tenant(s) with moderate load can use same database but different schema (approach in multenancy-op2 project).
- Tenant(s) with lower loads can share the DB SCHEMA (TENANT2 and TENANT3).
Tenant wise DB and Schema mapping.
| TENANTID | DB | SCHEMA |
|---|---|---|
| TENANT1 | dbA | SCHEMA1 |
| TENANT2 | dbB | SHARED |
| TENANT3 | dbB | SHARED |
cd multitenancy-op3
mvn clean package
mvn spring-boot:run
#####Create the following schema manually
create schema if not exists SHARED;
create table SHARED.catalogtable(id bigint, tenantid varchar(10), catalogname varchar(30), supplierid varchar(30), source varchar(10));
CREATE SEQUENCE SHARED.hibernate_sequence START WITH 1 INCREMENT BY 1 NO MINVALUE NO MAXVALUE CACHE 1;
Catalog creation request for TENANT1 which would insert data into table SCHEMA1.CATALOGTABLE in database dbA
curl -X POST http://localhost:8080/catalog -H 'content-type: application/json' -H 'tenantid: TENANT1' -d '{"tenantId":"TENANT1","catalogName":"Catalog1","supplierId":"supplier1","source":"X-Systems"}'
Catalog creation request for TENANT2 which would insert data into table SHARED.CATALOGTABLE in database dbB B
curl -X POST http://localhost:8080/catalog -H 'content-type: application/json' -H 'tenantid: TENANT2' -d '{"tenantId":"TENANT2","catalogName":"Catalog2","supplierId":"supplier2","source":"X-Systems"}'
Catalog creation request for TENANT3 which would insert data into table SHARED.CATALOGTABLE in database dbB
curl -X POST http://localhost:8080/catalog -H 'content-type: application/json' -H 'tenantid: TENANT3' -d '{"tenantId":"TENANT3","catalogName":"Catalog3","supplierId":"supplier3","source":"Y-Systems"}'
Inserted records can be verified at http://localhost:8080/h2-console/ by logging into use database dbA (jdbc:h2:mem:dbA) for TENANT1 in schema SCHEMA1
use database dbB (jdbc:h2:mem:dbB) for TENANT2 and TENANT3 in schema SHARED
curl -X GET 'http://localhost:8080/catalog?catalogId=1' -H 'content-type: application/json' -H 'tenantid: TENANT1'
Catalog retrival reqeust for TENANT2
curl -X GET 'http://localhost:8080/catalog?catalogId=1' -H 'content-type: application/json' -H 'tenantid: TENANT2'
Catalog retrival reqeust for TENANT3
curl -X GET 'http://localhost:8080/catalog?catalogId=2' -H 'content-type: application/json' -H 'tenantid: TENANT3'