How to use orchestral/tenanti in Laravel 5 to build a multi tenant application with multiple databases?

Question

I am trying to build and application using Laravel 5. It is supposed to be a multi tenant database architecture using multiple databases. My employer requires this for security

Answer 1

I've never used this package, but using the code you submitted above here's what I think is probably close to the right solution. You will probably still need to play with some of these values to get them correct:

Migration Paths

Since you're using the multi-database configuration, I believe you should be able to keep your migrations in the normal location, i.e. database/migrations. Tenanti will then create an exact replica of the database for each tenant in a different database. However, when you run php artisan tenanti:install user it might actually create a folder under database/ that indicates where you should put your migrations.

What is a "driver"?

The driver describes whether Tenanti will use a single or multiple databases, what models to use for determining different tenants, and where to store migrations. It is what you identified in the Tenanti config file you used above.

Database Connection Selection

You need to update config/database.php as follows. In a normal Laravel app, you would have the DB connection setup as follows:

<?php
    return [
        'fetch' => PDO::FETCH_CLASS,
        'default' => env('DB_CONNECTION', 'mysql'),
        'connections' => [
            'sqlite' => [ ...DB connection info... ],
            'mysql'  => [ ...DB connection info... ],
            'pgsql'  => [ ...DB connection info... ],
            'sqlsrv' => [ ...DB connection info... ],
        ],
        'migrations' => 'migrations',
        'redis' => [ ... ],
    ];

However, in the case of Tenanti multi-database setup, you need to add in different connection info for each tenant's database. To do this you would add a new level to your database.php config file (this example assumes you're using mysql, but you could use any DB, or even different database engines for different tenants):

<?php
    return [
        'fetch' => PDO::FETCH_CLASS,
        'default' => env('DB_CONNECTION', 'mysql'),
        'connections' => [
            'tenants' => [
                'user_1' => [
                    'driver'    => 'mysql',
                    'host'      => 'dbhost',     // for user with id=1
                    'database'  => 'dbname',     // for user with id=1
                    'username'  => 'dbusername', // for user with id=1
                    'password'  => 'dbpassword', // for user with id=1
                    'charset'   => 'utf8',
                    'collation' => 'utf8_unicode_ci',
                    'prefix'    => '',
                    'strict'    => false,
                ],
                'user_2' => [
                    'driver'    => 'mysql',
                    'host'      => 'dbhost',     // for user with id=2
                    'database'  => 'dbname',     // for user with id=2
                    'username'  => 'dbusername', // for user with id=2
                    'password'  => 'dbpassword', // for user with id=2
                    'charset'   => 'utf8',
                    'collation' => 'utf8_unicode_ci',
                    'prefix'    => '',
                    'strict'    => false,
                ],
           ],
        ],
        'migrations' => 'migrations',
        'redis' => [ ... ],
    ];

As you can see, each tenant has its own database instance that can be located on a different host and have a different username/password. Tenanti needs to be told how to figure out which database to use. This is what the documentation on Database Connection Resolver describes. In their example, they've named their tenant databases using acme_{$user->id} whereas in my example above I used user_{$user->id}.

Like I said, I've never actually set this up myself, but these are my best guesses based on the docs, and having used other packages by this same developer. Hope this helps!

Answer 2

+1 to @morphatic answer, it quiet accurate on most of the stuff.

Migration

One set of files is for the main DB which will store all the tenant information and the other files will be for the tenant DB. So how and where will these be stored?

For your main database you should be able to use the default database/migration and utilize php artisan make:migration and php artisan migrate.

Tenanti however will use the migration path set under the "driver" configuration. e.g:

'path' => database_path('tenanti/user'),

In this case the migration will be created/migrated from database/tenanti/user (you can choose other folder and it will use that folder). Once you set this up you can create new migration file for the user tenant via php artisan tenanti:make user create_blogs_table (as an example) and run migration via php artisan tenanti:migrate user (see the similarity between Laravel migration command and Tenanti?).

Driver

Driver is just the grouping of a tenant, you maybe grouping it by users, companies, or team etc. And there is possibility that you may require more than one type of group per project, otherwise most of the time you only be using single "group" or "driver".

Authentication or Accessing DB

How will I handle the authentication for the application? I mean whenever a tenant logs in, I will have to make sure the connection to the database changes dynamically. How will I accomplish this?

First of all, you need to consider how you're planning to distinguish each tenant. Most of the time I would see people tend to opt for subdomain. So in this case you need to check if the subdomain belongs to any of the user (by querying the main database) using a middleware and then connect to the database that belongs to the user.

Tenanti doesn't manage that part of the process, because everyone has different style on that aspect, but we do provide a code to dynamically connect to your database tenant from a base database configuration.

Let say you have the following config:

<?php

return [
    'fetch' => PDO::FETCH_CLASS,
    'default' => 'primary',
    'connections' => [
        'primary' => [
            //
        ],
        'tenants' => [
                'driver'    => 'mysql',
                'host'      => 'dbhost',     // for user with id=1
                'username'  => 'dbusername', // for user with id=1
                'password'  => 'dbpassword', // for user with id=1
                'charset'   => 'utf8',
                'collation' => 'utf8_unicode_ci',
                'prefix'    => '',
                'strict'    => false,
            ],
       ],
    ],
    'migrations' => 'migrations',
    'redis' => [ ... ],
];

You can follow the step available in https://github.com/orchestral/tenanti#multi-database-connection-setup and add the following code.

<?php namespace App\Providers;

use Orchestra\Support\Facades\Tenanti;

class AppServiceProvider extends ServiceProvider
{
    public function boot()
    {
        Tenanti::setupMultiDatabase('tenants', function (User $entity, array $template) {
            $template['database'] = "tenant_{$entity->getKey()}";

            return $template;
        });
    }
}

This would ensure that you be using tenant_1 database for user=1, tenant_2 database for user=2 and so on.

So how does Tenanti detect which user if active?

This is where you need to add logic in your middleware.

$user = App\User::whereSubdomain($request->route()->parameter('tenant'))->first();

Tenanti::driver('user')->asDefaultDatabase($user, 'tenants_{id}');