When the MobiLink server receives an upload from a MobiLink client, the entire upload is stored until the synchronization is complete. This is done for the following reasons:
Filtering download rows The most common technique for determining which rows to download is to download rows that have been modified since the most recent download. When synchronizing, the upload precedes the download. Any rows that are inserted or updated during the upload are rows that have been modified since the previous download.
It would be difficult to write a download_cursor script that omits from the download rows that were sent as part of the upload. For this reason, the MobiLink server automatically removes these rows from the download.
Processing inserts and updates By default, tables in the upload are applied to the consolidated database in an order that avoids referential integrity violations. The tables in the upload are ordered based on foreign key relationships. For example, if table A and table C both have foreign keys that reference a primary key column in B, then inserts and updates for table B rows are uploaded first.
Processing deletes after inserts and updates Deletes are applied to the consolidated database after all inserts and updates are applied. When deletes are applied, tables are processed in the opposite order from the way they appear in the upload. When a row being deleted references a row in another table that is also being deleted, this order of operations ensures that the referencing row is deleted before the referenced row is deleted.
Deadlock When an upload is being applied to the consolidated database, it may encounter deadlock due to concurrency with other transactions. These transactions might be upload transactions from other MobiLink server database connections, or transactions from other applications using the consolidated database. When an upload transaction is deadlocked, it is rolled back and the MobiLink server automatically starts applying the upload again, from the beginning.
Performance tip It is important to write your synchronization scripts to avoid contention as much as possible. Contention has a significant impact on performance when multiple users are synchronizing simultaneously.
Discuss this page in DocCommentXchange.
|
Copyright © 2012, iAnywhere Solutions, Inc. - SQL Anywhere 12.0.1 |