feat: AWS RDS Limitless Router Balancing #185
Description
Describe the feature
Provide the same ability to load balance across RDS Limitless routers as the JDBC driver does.
Use Case
I'm doing some testing for a SaaS application.
- I load two batches of 100k rows, to two different shardIds (1,2)
- Each batch is initiated in parallel with the other (kicked off with no await)
- Each batch is allocated 6 threads
- Each batch is microbatched into 1000 records with each of the 6 threads being assigned a microbatch
- Each microbatch is inserted using EFCore.BulkExtensions.MIT bulk insert
- Each microbatch gets a new dbContext and thus a new connection from the connection pool.
The behavior that I am seeing is that the insertion time for two batches is twice the time (plus a little more) than it is to insert one batch.
The theory is that this is due the connections being assigned to the same router. It's odd to me that the shard group doesn't handle this and it requires "special" awareness on client. If the client can figure this out, why can't the shardgroup?
Proposed Solution
Do it nearly identical to the AWS JDBC advanced driver.
Other Information
No response
Acknowledgements
- I may be able to implement this feature request
- This feature might incur a breaking change
The AWS Advanced .NET Data Provider Wrapper version used
n/a
dotnet version used
10.0.0
Operating System and version
macOS 15.7.2