Announcement: SnowMirror 3.0 Released

The SnowMirror team proudly announces the general availability of SnowMirror 3.0. We started developing SnowMirror 3 in cooperation with ServiceNow database experts several months ago. The cooperation resulted in a completely re-worked replication algorithm together with several new features improving SnowMirror’s performance and reliability. We do recommend upgrading the test SnowMirror installation first and then, if successful, upgrading the production mirrors afterward. By introducing indexes on the ServiceNow side the performance improves even more.

What Is New?
The replication algorithm no longer uses the sys_created_on column but only the sys_updated_on column. ServiceNow confirms that using this single column is better for overall performance. The disadvantage is that SnowMirror is not able to count the number of new records before the synchronization begins. It can only count created and updated records together. The pagination algorithm has been improved as well to be more reliable especially for large tables and tables under a load during the synchronization run. Some of the users were experiencing data losses and inconsistencies. These issues should be resolved by this improvement.

More features and fixes:
Synchronization validation – including the check of existing indexes
Consistency check – at the end of each run
Encoded queries and views replication improved
Synchronization filter and sorting – UI improvement
ServiceNow server time usage – requires Aggregate Web Service plugin
Clean & Synchronize using temporary tables
SnowMirror Monitor – Fuji-compatible ServiceNow application, the old update set is deprecated

More detailed information in the Release Notes and in the post covering the beta-testing.

IMPORTANT – Indexes
By using the new algorithm ServiceNow strongly recommends creating indexes on the ServiceNow side for larger tables to avoid full-table scans during the incremental loads. So please, create indexes on the sys_updated_on column for tables larger than 100k records. Either raise a ticket in the HI instance or, starting with the Fuji release, you can create an index on your own. Please read this article on how to introduce an index in Fuji.