Inflexible, but fully-managed
Updated September 12, 2018
Inflexible, but fully-managed
Score 7 out of 10
Vetted Review
Verified User
Overall Satisfaction with Stitch
We use Stitch to periodically replicate transactional datastores to an analytics warehouse. The transactional datastore are primarily PostgreSQL and MongoDB databases for microservices. Both the engineering and data teams use Stitch. The transactional datastores are not suitable for running analytics queries; they are tuned for different types of queries and access patterns. Extracting data from the transactional stores, transforming it to a different schema and data types, and loading it into a different datastore is not trivial work, and it is necessary to support the data and BI teams.
- Stitch is good at replicating relational databases. It manages the conversion of data types and schemas, and it is capable of incremental replication.
- Stitch's UI makes it easy to configure which tables should be replicated, and how they should be replicated.
- Stitch is not good at replicating document stores like MongoDB to relational databases. To be fair, this is a difficult task. Stitch flattens the objects, but the result is unwieldy.
- Stitch cannot replicate the same source to multiple sinks, which is inconvenient if you want to replicate some of a datastore's tables to Redshift and others to Redshift Spectrum, for instance.
- Stitch manages replication of transactional datastores to our data warehouse. Our data and business intelligence teams need this data to be replicated. Stitch allows our data engineers to focus on our product.
Stitch is less flexible than AWS DMS, but it is fully-managed.
Using Stitch
4 - Our organization has Stitch users on the data science, engineering, and business intelligence teams.