Bulk Sync

During the initial sync process in your application, you may need to deliver a large number of objects. This may result in many megabytes of data being sent to the device. In this case, incremental synchronization may take a while to complete since it uses JSON as the wire format. To significantly speed up this process you may use RhoConnect’s “bulk sync” feature.

• Bulk Sync Process
• Setup
• Data Files
• Bulk Sync Associations
• Using Bulk Sync with Fixed Schema Models
• Select which models to bulk sync.

Bulk Sync Process

If and when the Rhodes application issues a request to get bulk data, the RhoConnect application will do following:

• Queue a resque background job which will run the bulk sync process.

• Execute the ‘login,'query’,‘logoff’ methods of all source adapters matching the requested data partition to get latest set of objects from the backend service

• Generate appropriate database files for the device - hsqldb for Blackberry and sqlite for the rest of them. RhoConnect will generate sqlite files for iOS, Android, and Windows Mobile clients and hsqldb files for Blackberry (if enabled).

• The Rhodes application will poll RhoConnect for the bulk sync job to finish. Once the job is finished, the files are generated and Rhodes will download and install them to the device.

• After bulk database is installed, Rhodes application will analyze sources for references to remote blob objects. Each remote blob object will be downloaded to device, and if download is successful, remote references will be updated to local ones.

Once the bulk data sync process is complete for a source, you may continue to synchronize sources using the default incremental sync process. However, the default sync process may be disabled for a source by setting its sync_type to ‘bulk_sync_only’. This will save space in your redis database if you don’t plan on using incremental sync for this source.

For example, your settings/settings.yml might have:

:sources:
  Product:
    poll_interval: 300
  Customer:
    poll_interval: 300
    sync_type: bulk_sync_only

Setup

Setup DB Library

Bulk sync prepares database files inside of a job, therefore it needs access to sqlite3’s ruby library. To enable this library, uncomment or add the following line to your RhoConnect app’s Gemfile:

gem 'sqlite3', ">= 1.3.3", :platforms => [:ruby, :mswin, :mingw]

Then install the gem:

$ bundle install

Enabling Bulk Sync in Rhodes

In the device rhodes application, setting bulksync_state to 0 in rhoconfig.txt will trigger bulk sync next time sync is performed:

bulksync_state = 0

After application executed bulk sync it will set bulksync_state to 1 and next time sync will happen, it will not execute bulk sync anymore. You may force bulk sync on the next sync cycle by setting bulksync_state to 0 again using ruby in your application:

Rho::RhoConfig.bulksync_state = '0'

Enabling BlackBerry Bulk Sync

If your application will be performing bulk sync for BlackBerry devices, you will need to install the open source Rhomobile HSQLDB tool hsqldata.jar in your RhoConnect application.

1. Download hsqldata.

2. Build and install hsqldata.jar into your RhoConnect application:

     $ cd hsqldata
    $ ant
    $ mkdir <path-to-rhoconnect-app>/vendor
    $ cp bin/hsqldata.jar <path-to-rhoconnect-app>/vendor/
   

3. Next set the :blackberry_bulk_sync: flag in settings/settings.yml:

     :development:
      :licensefile: settings/license.key
      :redis: localhost:6379
      :syncserver: http://localhost:9292/application/
      :blackberry_bulk_sync: true
   

Data Files

By default, bulk sync data files will be stored in the application’s root folder, under a ‘data’ directory. When bulk data is requested by the Rhodes application, RhoConnect will return a link to the file if the bulk_sync_poll_interval has not expired. If the file is expired, RhoConnect will return a wait command to Rhodes and enqueue another resque job to generate new bulk data files. By default, the bulk_sync_poll_interval is set to 1 hour (3600) seconds and is configurable in settings.yml:

:development: 
  :redis: localhost:6379
  :licensefile: settings/license.key
  :syncserver: http://localhost:9292/application/
  :bulk_sync_poll_interval: 3600

Bulk Sync Associations

Sometimes your bulk data is not simply a collection of separate sources, but they are linked together through associations. For these purposes RhoConnect uses the notation of belongs_to association. The belongs_to association is a natural way to express hasOne and hasMany associations. In the case of bulk sync, data belongs_to relationship includes a list of all dependent objects for model(s) and helps resolve dependency issues during bulk sync process.

Associations are defined in your RhoConnect application’s settings/settings.yml file.

For example, you can have a list of customers who have purchased a product, and thus you can have them belong to that product:

:sources:
  :Product:
    :poll_interval: 600
    :partition_type: app
    :sync_type: bulk_sync_only     
  :Customer:
    :poll_interval: 300
    # ...
    belongs_to:
      - :product_id: 'Product'

belongs_to format is an array of attribute/source_name pairs. It can be polymorphic and define associations across multiple sources:

  belongs_to:
    - :parent_id, 'Product'
    - :parent_id, 'Cases'

Bulk Sync Associations are similar to Rhom Sync Associations and both should be considered together if you are planning to use bulk sync data.

Using Bulk Sync with Fixed Schema Models

If your Rhom model is a fixed schema model and you use bulk sync, you can control the bulk sync schema at runtime in your source adapter by implementing the ‘schema’ method.

An example schema method is as follows:

def schema
  {
    'version' => '1.0',
    'property' => {
      'name' => 'string',
      'brand' => 'string',
      'price' => 'string',
      'image_url_cropped' => 'blob,overwrite',
      'image_url' => 'blob'
    },
    'index' => {
      'by_name_brand' => 'name,brand'
    },
    'unique_index' => {
      'by_price' => 'price'
    }
  }.to_json
end

Select which models to bulk sync.

Bulk sync process will only sync sources which sync type is not ‘none’. You may exclude model data from bulk sync with following code in your Rhodes application:

Rho::RHO.get_user_db().update_into_table('sources',{'sync_type'=>'none'}, {'name'=> "MySourceName" })

Set sync type to ‘incremental’ or ‘bulk’ when you want to get model data with bulk sync.