Skip to main content

Overview

Clustron DKV is not just a key-value store — it is a distributed coordination system built around a simple data model.

To use Clustron effectively, it helps to understand a few core ideas.

The Mental Model​

At its core, Clustron provides:

  • A distributed key-value store
  • A set of coordination primitives
  • A consistent programming model across local and distributed environments

You can think of Clustron as:

Shared state + coordination + lifecycle management

The Store​

A store is the primary unit in Clustron.

  • It represents a logical cluster

  • It contains data and coordination primitives

  • It can run:

    • In-process (single application)
    • Distributed (multiple nodes across machines)

All operations in Clustron are performed against a store.

Data + Coordination​

Clustron combines two things that are usually separate:

Data (Key-Value)​

  • Store and retrieve values using keys
  • Apply expiration using TTL
  • Use transactions for multi-step operations

Coordination​

  • Locks → ensure mutual exclusion
  • Counters → track shared state
  • Watch → react to changes in real time

These primitives allow multiple application instances to work together safely.

Lifecycle Awareness​

Unlike traditional caches, Clustron understands lifecycle:

  • Keys can expire automatically
  • Resources can be bound to leases
  • Failures are handled through expiry and ownership

This makes Clustron suitable for building reliable distributed systems.

One Model, Two Modes​

Clustron uses the same API in both modes:

In-Process​

  • Runs inside your application
  • No infrastructure required

Distributed​

  • Runs across multiple nodes
  • Enables coordination between instances

You can move from local to distributed without changing your code.

How Everything Fits Together​

Store
  ├── Keys & Values
  ├── TTL & Expiration
  ├── Leases (lifecycle control)
  └── Coordination (locks, counters, watch)

Why This Matters​

Clustron is designed to solve problems like:

  • Coordinating background workers
  • Managing distributed locks
  • Tracking shared state across instances
  • Reacting to changes in real time

Instead of combining multiple systems, you can use a single, consistent model.

What’s Next​

👉 Continue to Architecture to understand the design of Clustron DKV.