In V12, add (restore) the capability to manage transaction logging at cube level.

Context

V12 relies on transaction logs, a feature that is equivalent to that implemented in V11, through a Mongo DB. This feature notably allows data to be shared between replicas.

While the transaction log is essential and justified for instance replication in V12, it is also crucial to provide a 'no replication' option for cases where replicas are not used. This would allow logging to be disabled at cube level, as in V11.

Justification

In our application, we have 30 regional data entry and simulation instances, consolidated into three central group reporting instances. During peak periods, these processes transfer up to 3 billion rows per day from the regional instances to the central consolidation instances.

Currently, transaction logs on these central instances are disabled in V11, for two main reasons:

• Performance, as these instances are fed exclusively by replayable files and are read-only for users,

• Infrastructure efficiency, as the volume of transaction logs would be substential, affecting both storage and processing speed.

Additionally, concerns have been raised about the performance of this component in V12 as the number of instances increases, since transaction logging is managed by the same MongoDB component for all instances in an environment, which is linked to the same pool of physical resources.

In V11, transaction logging is managed per host – i.e. per Windows server - so adding a new TM1 instance on a new Windows server has no impact on the others.