Skip to main content

Consistency Model

Clustron DKV provides a consistent and predictable view of data across nodes.

It is designed to balance:

  • Correctness
  • Availability
  • Performance

What Does Consistency Mean?​

Consistency defines how data behaves when:

  • Multiple nodes are involved
  • Multiple clients access the same key
  • Updates happen concurrently

Single Logical Store​

Even in distributed mode, Clustron behaves as a single logical store:

Multiple nodes → One consistent view

Clients do not need to manage individual nodes.

Read and Write Behavior​

Writes​

  • Writes are applied through the cluster
  • The system ensures a consistent state is maintained

Reads​

  • Reads return the latest known value
  • After a successful write, subsequent reads will reflect the updated value

Consistency in Practice​

For most applications, Clustron provides:

  • Predictable reads and writes
  • No need to manually handle replication
  • No need to manage consistency across nodes

You interact with the system as if it were a single instance.

Concurrent Updates​

When multiple clients update the same key:

  • Operations are handled safely
  • The system ensures a consistent outcome

For coordination scenarios, use:

  • Locks → enforce exclusive access
  • Leases → control ownership and lifecycle

Consistency Across Modes​

The same consistency model applies to:

  • InProc mode
  • Distributed mode

Your application does not need different logic.

Design Philosophy​

Clustron aims to provide:

Simple mental model  
+  
Reliable behavior  
+  
No manual coordination

When You Need More Control​

For advanced scenarios, you can use:

  • Locks → prevent conflicting updates
  • Transactions → group multiple operations
  • Leases → manage ownership and expiry

What’s Next​

👉 Continue to Leases to understand lifecycle and ownership