Bounded staleness is frequently chosen by globally distributed applications that expect low write latencies but require total global order guarantee. Bounded staleness is great for applications featuring group collaboration and sharing, stock ticker, publish-subscribe/queueing etc.
Scenario: Application updates in con-product frequently cause HTTP status code 429 "Too many requests". You discover that the 429 status code relates to excessive request unit (RU) consumption during the updates.
Reference:
https://docs.microsoft.com/en-us/azure/cosmos-db/consistency-levels

Based on the characteristics of the application and the provided document structure, the most suitable partition key value for coll1 in the given scenario would be the customerId, Option B.
The application frequently creates new orders by different customers and customers often view their past order history. Using customerId as the partition key would ensure that all orders associated with a particular customer are stored in the same partition. This enables efficient querying of past order history for a specific customer and reduces cross-partition queries, resulting in lower costs and improved performance.
a partition key is a JSON property (or path) within your documents that is used by Azure Cosmos DB to distribute data among multiple partitions3. A partition key should have a high cardinality, which means it should have many distinct values, such as hundreds or thousands1. A partition key should also align with the most common query patterns of your application, so that you can efficiently retrieve data by using the partition key value1.
Based on these criteria, one possible partition key that you could use for coll1 is B. customerId.
This partition key has the following advantages:
It has a high cardinality, as each customer will have a unique ID3.
It aligns with the query patterns of the application, as customers will often view their past order history3.
It minimizes costs, as it reduces the number of cross-partition queries and optimizes the storage and throughput utilization1.
This partition key also has some limitations, such as:
It may not be optimal for scenarios where orders need to be queried independently from customers or aggregated by date or other criteria3.
It may result in hot partitions or throttling if some customers create orders more frequently than others or have more data than others1.
It may not support transactions across multiple customers, as transactions are scoped to a single logical partition2.
Depending on your specific use case and requirements, you may need to adjust this partition key or choose a different one. For example, you could use a synthetic partition key that concatenates multiple properties of an item2, or you could use a partition key with a random or pre-calculated suffix to distribute the workload more evenly2.