能否不依赖DynamoDB使用AWS AppSync？拟改用MongoDB后端

Absolutely, you don’t have to use DynamoDB with Amazon AppSync—this is one of the platform’s most flexible features. And yes, you can absolutely pair it with MongoDB while retaining full access to AppSync’s offline and sync capabilities. Let me break this down with practical context:

Why Ditching DynamoDB (and VTL) for MongoDB Works #

I get it—writing VTL resolvers for DynamoDB can feel clunky, especially if you’re more comfortable with MongoDB’s query patterns. AppSync supports multiple data source types, so you can bypass DynamoDB entirely by using either:

• HTTP Data Sources: Directly connect to a MongoDB Atlas endpoint (or your self-hosted MongoDB instance) via a REST API wrapper.
• Lambda Data Sources: Use a Lambda function as an intermediary to handle all interactions with MongoDB. This is the most popular approach because it lets you write logic in languages like Node.js, Python, or Java—no more VTL headaches.

Offline Sync Compatibility #

AppSync’s offline functionality is built into its client SDKs (iOS, Android, JavaScript) and relies on local caching + conflict resolution logic, not the underlying backend. As long as your MongoDB-backed resolvers handle the following, offline sync will work seamlessly:

• Implement versioning or timestamps on your MongoDB documents to support AppSync’s conflict resolution strategies (like optimistic locking).
• Ensure your resolvers correctly return data structures that match your GraphQL schema, so the client cache can sync changes properly when back online.
• Handle create/update/delete operations in a way that aligns with AppSync’s sync workflow (e.g., returning the updated document after a mutation, so the client can reconcile local changes).

A Quick Implementation Example #

Here’s a high-level workflow for setting up MongoDB with AppSync:

1. Define your GraphQL schema in AppSync (queries, mutations, types that map to MongoDB documents).
2. Create a Lambda function that connects to your MongoDB instance (using drivers like mongoose for Node.js).
3. Configure AppSync resolvers to point to this Lambda function for all schema operations—no VTL required.
4. In your client app, enable AppSync’s offline mode (via SDK settings like enableOfflineMutations: true).
5. Add version fields (e.g., _version) to your MongoDB documents to let AppSync handle conflict detection during sync.

Key Considerations #

• Performance: Using Lambda as an intermediary adds a small latency layer, but for most use cases, this is negligible. You can optimize by keeping Lambda functions close to your MongoDB instance and using connection pooling.
• Conflict Resolution: Make sure your backend enforces version checks—if a client tries to sync a change based on an outdated document version, your Lambda should reject it, and AppSync will prompt the client to resolve the conflict.
• Security: Use IAM roles to secure AppSync access to your Lambda, and ensure your MongoDB instance is only accessible via private networks or authenticated endpoints.

To sum it up: You absolutely can use MongoDB with AppSync while keeping all the offline sync features you need, and it’s a great way to avoid the pain of writing DynamoDB-focused VTL resolvers.

内容的提问来源于stack exchange，提问作者honkskillet