Aria Jelinek outlines the value of Azure Synapse Link:
New as of Ignite 2021, customers can optimize queries by setting custom partitions for their Azure Cosmos DB analytical store using keys that are commonly used as query filters. This compacts and optimizes the analytical data written to the partitioned store, resulting in better query performance even when working with a high volume of update or delete operations.
Azure Synapse Link is also now available for Azure Cosmos DB serverless accounts, expanding the integration to cover data from workloads with bursts of traffic or uncertain traffic patterns.
This post mostly covers the Dataverse and Cosmos DB integrations rather than the integration with SQL Server 2022.
One the whole, I like Azure Synapse Link for Cosmos DB and will probably like it for SQL Server 2022—maybe even a bit more. It does simplify the ELT process by taking care of the E and handling the first half of the L (landing into a staging table). Though if data’s going into a dedicated SQL pool, I do hope the people doing this understand that dedicated SQL pools are intended for Kimball-style data warehousing scenarios and there can be a considerable performance (and therefore price) hit if you simply replicate a bunch of stuff without subsequent transformation.