Amazon Dynamo minutes

1. Introduction
2. Background
2.1 System Assumptions and Requirements
2.2 Service Level Agreements(SLA)
2.3 Design consideration

3. Related Work
3.1 Peer to Peer System
3.2 Distributed File System and Database
3.3 Dscussion

4.System architectue

4.1 System interface

4.2 Partitioning Algorithm

4.3 Replication 

-------------------------------

Problem: Partitioning
Technique: Consistent Hashing
Advantage: Incremental Scalability

Problem: High Availability for writes
Technique:Vector clocks with reconciliation during reads
Advantage: Version size is decoupled from updated rates

Problem: Handling temporary failures
Technique: sloppy quorum and hinted handoff
Advantage: Provides high availability and durability guarantee when some of the replicas are not available

Problem: Recovering from permanent failures
Technique: Anti-entropy using Merkle trees
Advantage: Synchronizes divergent replicas in the background

Problem: membership and failure detection
Technique: Gossip-based membership protocol and failure detection
Advantage: Preserves symmetry and avoids having a centralized registry for storing membership and node liveness information. 

 

(To be continued...)