Building Content-Based Publish/Subscribe Systems with Distributed Hash Tables

Tam, Azimi, Jacobsen

publish/subscribe pub/sub dht distributed hashtables

@inproceedings{tam:disp2pc-2003,
  title={Building Content-Based Publish/Subscribe Systems
         with Distributed Hash Tables},
  author={Tam, D. and Azimi, R. and Jacobsen, H.A.},
  booktitle={International Workshop on Databases, Information Systems,
             and Peer-to-Peer Computing},
  pages={138--152},
  year={2003},
  publisher={Springer-Verlag}
}

Attach:Tam-DISP2PC2003.pdf

Automatically categorizes content into topics based on a template schema

• All nodes must have the schemas
• Schemas stipulate index variables in the schema, e.g., price, brand
• Publications and subscriptions are compared to the schemas and any applicable indexes extracted
• These produce digests, which are then used to hash the request into the DHT

Subscriptions are stored at each of the nodes in the DHT path from the subscriber to the root (key owner)

Publications are forwarded to the root, and then propagated down the tree

Range queries are difficult to do over DHTs

• Addressed here by breaking into standard intervals

Of note:

• Scribe topic based pub/sub
  • Rendezvous at nodes mapped from topic key
• SIENA
• SkipNet