Almost all TimesTen In-Memory Database customers use some form of high availability configuration as they want high performance RDBMS solutions without a single point of failure.
Not all customers have the same sort of applications or availability requirements, so they have to choose which TimesTen replication configuration best suits them.
The following picture shows a subset of the possible replication configurations:
The following picture summarizes the main differences between Active Standby Pair based and Classic Replication:
Active Standby Pair replication schemes can also optionally be configured to include a set of read only subscriber databses:
For 'classic replication' (ie non active standby pair configurations) there are several options.
Another thing to consider for replication configurations is whether the data is sent synchronously or asynchronously.
If you need to support read/write caching for Oracle 11g/12c data with TimesTen Application Tier Database Cache, then you have to use one of the Active Standby Pair configurations. Active Standby Pair replication configurations are the simplest to understand, configure and manage.
The split workload and distributed workload configurations for Classic Replication give more flexibility, but careful consideration is needed to avoid and/or mitigate write/write conflicts.
The active standby pair and classic replication configurations are building blocks that can be used for scale out architectures. For example, some TimesTen telecommunications customers with high throughput requirements, use application level partitioning to route transactions to a set of active standby pairs.
Disclaimer: these are my personal thoughts and do not represent Oracle's official viewpoint in any way, shape, or form.