At a high level, Dynamo is a

that is replicated across the cluster for high availability and .

Introduction: Dynamo's architecture

Dynamo's architecture can be summarized as follows (we will discuss all of these concepts in detail in the following lessons):

1. Data distribution

Dynamo uses Consistent Hashing to distribute its data among nodes. Consistent hashing also makes it easy to add or remove nodes from a Dynamo cluster.

2. Data replication and consistency

Data is replicated optimistically, i.e., Dynamo provides

.

3. Handling temporary failures

To handle temporary failures, Dynamo replicates data to a sloppy quorum of other nodes in the system instead of a strict majority

.

4. Inter-node communication and failure detection

Dynamo's nodes use gossip protocol to keep track of the cluster state.

5. High availability

Dynamo makes the system "always writeable" (or highly available) by using hinted handoff.

6. Conflict resolution and handling permanent failures

Since there are no write-time guarantees that nodes agree on values, Dynamo resolves potential conflicts using other mechanisms:

• Use vector clocks to keep track of value history and reconcile divergent histories at read time.
• In the background, dynamo uses an anti-entropy mechanism like Merkle trees to handle permanent failures.

Let's discuss each of these concepts one by one.