Salesforce automatically indexes primary keys.

According to the Salesforce documentation, a hub-and-spoke model is an integration architecture pattern that allows connecting multiple Salesforce orgs using a central org (hub) and one or more secondary orgs (spokes). The hub org acts as the master data source and orchestrates the data flow between the spoke orgs. The spoke orgs act as the consumers or producers of the data and communicate with the hub org.
To orchestrate data between the hub org and spoke orgs, a data architect should use:
A middleware solution that extracts and distributes data across both the hub and spokes (option A). This means using an external service or tool that can connect to multiple Salesforce orgs using APIs or connectors, and perform data extraction, transformation, and distribution operations between the hub and spoke orgs. This can provide a scalable, flexible, and reliable way to orchestrate data across multiple orgs.

Leveraging Heroku Connect and Heroku Postgres, middleware tools, or Salesforce Connect and the cross-org adapter are all viable options to share data between Org A and Org B in near-real time. Installing a 3rd party AppExchange tool may not provide bidirectional sync or may have additional costs. Developing an Apex class that pushes opportunity data between orgs daily via the Apex schedule may not meet the near-real time requirement.

Archiving and purging old data from Salesforce on a monthly basis is a good strategy to improve the user experience for the Sales Reps, as it will reduce the clutter and improve the search performance. Creating a permission set or setting data access to private are not effective ways to hide old data from Sales Reps, as they will still consume data storage and affect search results.
Using Batch Apex to archive old data on a rolling nightly basis is also not a good option, as it will consume API requests and processing time, and may not comply with the data retention policy.