Delta CRDTs: Bandwidth‑Efficient Edge Collaboration

Scenario context

🚀 Delta CRDTs let edge devices sync only the changes, slashing bandwidth and latency—crucial for offline‑first, low‑latency collaboration. Edge devices (smartphones, IoT sensors, remote terminals) often have intermittent connectivity and limited uplink bandwidth. When a CRDT (Conflict‑Free Replicated Data Type) is used naively, every update pushes the entire data structure to peers. In a 10‑kB counter that grows…

This walkthrough focuses on The Gotcha.

Before making a design choice, clarify:

• Edge devices (smartphones, IoT sensors, remote terminals) often have intermittent connectivity and limited uplink bandwidth.
• When a CRDT (Conflict‑Free Replicated Data Type) is used naively, every update pushes the entire data structure to peers.
• In a 10‑kB counter that grows to 1 MB after a week of edits, this becomes a nightmare: high latency, battery drain, and data caps.

Concepts in play: consistency-replication, jvm-internals-and-garbage-collection

Stage 1 — The Solution – Delta CRDTs

• Delta CRDTs store only the change (delta) that occurred, not the whole state.
• Each operation produces a small delta object that is merged into the local state and propagated to peers.
• When a peer receives a delta, it merges it into its state and can optionally send back a compressed acknowledgment.
• Because deltas are usually a few bytes, bandwidth usage drops dramatically, and the time to apply changes remains low.

Observable outcome: Because deltas are usually a few bytes, bandwidth usage drops dramatically, and the time to apply changes remains low.

Stage 2 — Code Snippet – GCounter Delta CRDT (TypeScript)

• When a device updates its counter, it calls deltaInc to get the delta and broadcasts it.
• Peers run mergeDelta on receipt.
• The full state can be reconstructed by summing all values in the map.

type Delta = { id: string; inc: number };
type GCounter = Map<string, number>;

function deltaInc(counter: GCounter, nodeId: string, n: number): Delta {
  const current = counter.get(nodeId) ?? 0;
  counter.set(nodeId, current + n);
  return { id: nodeId, inc: n }; // delta to send
}

function mergeDelta(counter: GCounter, delta: Delta): void {
  const current = counter.get(delta.id) ?? 0;
  counter.set(delta.id, current + delta.inc); // idempotent merge
}

Observable outcome: The full state can be reconstructed by summing all values in the map.

Stage 3 — Architecture Kit – Delta Propagation Flow

• { "version": 1, "title": "Delta CRDT Sync at the Edge", "description": "Local updates generate deltas; peers exchange deltas over unreliable links.", "view": "data-flow", "pattern": "gossip", "direction": "LR", "groups"…

Observable outcome: { "version": 1, "title": "Delta CRDT Sync at the Edge", "description": "Local updates generate deltas; peers exchange deltas over unreliable links.", "view": "data-flow", "pattern": "gossip", "direction": "LR", "groups"…

End-to-end outcome

After working through the stages, you should be able to explain:

• Delta CRDTs are a practical pattern for edge‑first collaborative apps: they keep network traffic minimal, preserve low‑latency updates, and maintain the strong guarantees of CRDTs.
• Use them whenever you need to stay responsive on bandwidth‑constrained, unreliable links.

Alternatives considered

The source focuses on one primary design. A second architecture is not compared, so no alternative is invented here.

Failure and degraded modes

The source does not isolate a separate failure mode. Limits that matter for the scenario are called out in the stage decisions above.

• Delta CRDTs are a practical pattern for edge‑first collaborative apps: they keep network traffic minimal, preserve low‑latency updates, and maintain the strong guarantees of CRDTs.
• Use them whenever you need to stay responsive on bandwidth‑constrained, unreliable links.
• Delta CRDTs store only the change (delta) that occurred, not the whole state.
• When a device updates its counter, it calls deltaInc to get the delta and broadcasts it.
• { "version": 1, "title": "Delta CRDT Sync at the Edge", "description": "Local updates generate deltas; peers exchange deltas over unreliable links.", "view": "data-flow", "pattern": "gossip", "direction": "LR", "groups"…