• User: guest
  • Change login
  • Register
  • Change locale

SData Linking & Synchronisation

This section describes the SData Link & Sync protocol.

The first subsections describe the linking protocol:

• Correlating with UUIDs
• Linking process
• Linking protocol
• Linking scenarios

The following subsections describe the synchronization algorithm, which is a variant of the well-known vector clock synchronization algorithm:

• Basic vector clock synchronization algorithm.
• Tick management
• Basic conflict handling
• "Digest" variant of the algorithm, which allows SData to efficiently synchronize collections of resources.
• Selecting changes with digest
• Conflict handing with digest

The following subsections describe the architecture of SData synchronization. They cover:

• Terminology
• Reference architecture
• Architecture variants
• Immediate and catch up synchronization modes.

Once the architecture has been laid out, we describe the protocol elements:

• Synchronization URLs
• Synchronization digest
• Per-resource synchronization state 
• Synchronzation feed
• Synchronization metadata.

Then, we dissect a typical synchronization pass. We show how the engine interacts with the URLs to:

• Read the target digest
• Query the source synchronization feed
• Update the target
• Handle conflicts on the target side
• Iterate through source pages 
• Log the results

We also give an example of immediate synchronization.

After this walkthrough, we describe advanced topics (not published yet): 

• Relationship handling
• Partial storage at endpoints 
• Field-level synchronization

Finally, we give some implementation hints (not published yet):

• Engine implementation notes
• Endpoint implementation notes

---