Right now we are using local SQL Server 2016 as the DB server and planning to move it to Amazon RDS. The issue is, currently there is a filestream filegroup in use to store some occasional binary blobs like some small images, few json and xml files, some MS Office Documents etc... The usage is very occasional, and the file sizes range from few hundred KBs to few MBs.
Since RDS SQL Server is not yet supporting FILESTREAM, it must be moved out.
So what is the ideal recommended approach? 1. Amazon DynamoDB - By creating a wrapper object to contain the documents. But the 400KB attribute restriction is a problem. 2. Amazon S3 as document storage, and putting the S3 url in SQL Server instead of the binary blob. 3. Continue using SQL Server, but move the FILESTREAM to a different rows data filegroup so that it will not interfere with primary row data.
Considering the programming difficulty, option#3 would be the easiest, since absolutely no coding is necessary and the change is limited to some schema changes; and option#2 would be the most difficult (since it requires proper programming pattern to be brought in).
What would be the best practice here, considering the programming difficulty, performance, scalability/availability and usage?
